Simulation savings
10 Jun 2015
The use of dynamic simulation software can radically reduce downstream build costs and cut upstream shutdown times.
Upstream oil & gas operators are just getting used to their profits being squeezed due to tumbling oil prices, but for their downstream counterparts tight margins and difficult market conditions have been a common theme over the past few years.
As a result, refineries and petrochemical plants have been forced to embrace all kinds of technology that can boost their efficiency.
And if there is a piece of technology that can shave millions off project costs, then they are understandably keen to embrace it.
These market conditions explain the rise of dynamic simulation being used for process modelling in both the design of new plants, as well as the redesign of existing sites for new feedstocks.
The key area where dynamic simulation has been used in downstream oil & gas has been in calculating distillation column relief loads as part of the flare system for a refinery or petrochemical plant.
Conventional methods to calculate column relief loads look at flare systems in their “steady-state” and are conservative. This can lead to flares that are over-designed and far more expensive than they need to be.
Flare design
French contractor Technip used Schneider’s SimSci Dynsim software to model the design of a new refinery flare system.
When comparing the results of dynamic simulation with more traditional methods, Technip found that it was able to save €20 million on just one flare header by safely reducing the capacity that it needed to be.
This saving equated to 32% of the entire flare system cost.
“What we have found is that overdesign can be between 4% and 80% - roughly four times bigger than it needs to be,” says Schneider Electric SimSci Dynsim product manager Gregor Fernholz.
“With refineries’ margins at 2-3%, obviously even at just a 20% overdesign [the business] would be making a loss.”
The efficiencies generated by dynamic simulation are due to the software enabling far more accurate design by assessing a range of scenarios that could impact on a flare system, rather than just its steady state.
For example, Flexible Process Consultants (Flex Process) senior simulation engineer Ben Firth says that when sizing a common flare header, one of the more difficult problems an engineer faces is calculating the load from a multiple relief event.
“A common approach is to assume each unit will relieve against 50% of the load from the other units, unless there is a known simultaneous event, in which case it is against the full loads,” says Firth.
“Further rules of thumb apply to taking credit for trips.”
But these rules of thumb bring up further questions, he adds.
“Are we confident we understand the timings of the multiple events?” asks Firth.
“Are the trips set properly to achieve their objectives in time? Do we fully understand the wider process effects of the trips? Does the trip on one unit exacerbate or mitigate the relief load on another?”
Building a model
When building a model, particularly of an existing plant, accuracy is the key, says Firth.
“The old adage of modelling, ‘Garbage in, garbage out’, always applies,” he says.
“We use every datasheet, general arrangement drawing and piping isometric, along with plant historical data, to match plant operation as closely as we can. Once we and the client are happy that the model is ready, we can begin a relief and flare study.”
Such a study involves going through every identified scenario, including common mode failures (e.g. utilities), and ensuring that every relief device, as well as the flare system, is appropriately sized.
Because of the real-time nature of the modelling, and Flex Process’ in-depth model rating, Firth says he can be confident that the timings and severities of the relief events modelled are reliable.
One example of simulating multiple relief events is that of a power trip at the now-closed Murco refinery in Milford Haven, Wales (see box below).
Firth says he believes that no engineer on their own could have modelled all of the knock-on events resulting from the power failure that were calculated by the dynamic simulation software.
“One may have been able to determine all of this from pumping and instrumentation diagrams and experience, but we think the average engineer would assume the Naphtha unit relieves, while missing the complex series of events affecting the Fuel Gas system,” he says.
“And even if one does catch all of this, what about the timings? Because it brings clarity to complex situations, and eliminates rules of thumb, we firmly believe that dynamic simulation represents best practice for relief and flare studies.”
Multifunctional Models
Besides helping make the plant safer, modelling provides a range of other benefits, adds Firth.
“Debottlenecking is an instant by-product, as plant inefficiencies are reproduced on screen,” he says.
“The models also support design, modifications, as well as control tuning, safety instrumented system design, and trip checkout. Additionally, connecting to control software provides an operator training simulator that matches the actual plant.”
This last benefit is one that also applies to upstream oil & gas.
Fernholz says that use of the simulation software ahead of the upgrade of equipment has enabled operators to more than halve their anticipated shutdowns times for the upgrades, as accurate modelling has ensured that the equipment worked first time.
Operators were also trained using simulation software to ensure they could use the new equipment straight away.
Simulating failure at Milford Haven refinery
Prior to its closure late last year, Flexible Process Consultants (Flex Process) used dynamic simulation to model various relief events at Murco’s refinery in Milford Haven.
“The site power came in via two 11kV transformers, dividing the power supply into Side A and Side B,” says Flex Process senior simulation engineer Ben Firth.
“One failure we examined was the loss of Side B, resulting in several relief events along a common flare header.”
During the first four minutes of simulating the loss of Side B, the Fuel Gas balance drum relieves with increasing frequency, as most of the units see a significant bump in off-gas, while furnaces wind down, or are tripped.
At four minutes, the Debutaniser overhead drum overfills and dumps flashing liquid propane gas (LPG) into the flare and the Fuel Gas system.
This greatly increases the Fuel Gas relief load. At seven minutes the Kerosene stripper relieves, as the feed becomes significantly hotter, swamping the overhead system.
This is followed by the VNS reflux drum overflowing at 10 minutes, due to loss of pumps.
This causes the VNS tower to relieve after 16 minutes, combining with its overhead liquid to produce slug flow.
The result is unacceptable backpressure on the still relieving Fuel Gas drum and Kerosene stripper. Meanwhile, two towers on the Naphtha unit threaten to relieve after losing overhead pumps.
However, a knock-out pot overfills, tripping the compressors, and the unit winds down.
“All of this looks bad, but hope is not lost,” says Firth.
“Understanding is the first step to solving a problem. First, we need to deal with the flashing LPG in the flare header. Installing a high level trip on the Debutaniser overhead drum achieves this.”
This also eliminates LPG in the Fuel Gas, which now stops relieving after 11 minutes, avoiding the backpressure issue at 16 minutes. Coolers on the Kerosene stripper feed can be wired into power Side A.
Finally, the slug flow can be dealt with using a new high level trip on the VNS reflux drum, and an existing pressure switch on the VNS tower.
“We also used the model’s slug flow data for a mechanical integrity study,” adds Firth.
“All of the recommendations were re-tested in the models. These simple, cost-effective solutions meant a legacy flare header could remain in service.”
