P. J. Parsons, J.J. Templeman
British Gas plc Midlands Research Station
Solihull, West Midlands U.K.
Models to predict performance of a natural-gas absorber for dew point control were used successfully to analyze absorber performance at British Gas plc's Easington terminal and have led to modifications to handle additional volumes.
The models were verified with data from pilot-plant tests and trials on absorbers at the terminal.
They revealed the absorbers' capacities to be greater than design and allowed for flowrates of twice the original design value to be processed.
DEW POINT SPECS
Natural gas must be treated to remove water and higher hydrocarbons before it is passed from the production wellhead into a high-pressure transmission system. This is to avoid operating problems that would result from the accumulation of liquids in the pipelines.
For gas produced from the U.K. southern sector of the North Sea, a minimum of processing is carried out off-shore. Fig. 1 shows a simplified processing scheme.
The removal of sand, free water, and hydrocarbon condensate is followed by either sufficient dehydration to prevent corrosion and hydrate formation in the marine pipeline or by injection of corrosion and hydrate formation inhibitors. Condensate separated offshore is reinjected into the marine pipeline.
At the onshore reception terminal, further processing of the gas is then required to meet the water and hydrocarbon dew point specifications for transmission.
British Gas' Rough field,' which uses a partially depleted gas reservoir, is the largest offshore gas-storage facility in the world. Fig. 2 shows one of the two offshore installations.
Natural gas from the U.K. National Transmission System (NTS) is injected into the field during the summer when demand is at its lowest and withdrawn in the winter at times of highest demand. The Rough field can supply up to 10% of the U.K.'s peak demand.
The gas is processed onshore at the Easington reception terminal in North Humberside.
In considering a process scheme for dew point control, it was anticipated that some mixing of the injected gas from the NTS and indigenous gas from the Rough field reservoir (Table 1) would take place during storage. Thus the process selected had to be designed for a range of gas mixtures.
Preliminary studies showed that certain mixtures containing a high proportion of injected gas could not be processed to meet the dew point specification with a conventional refrigeration process operating at - 23 C. and 70 bar (1,015 psi) because insufficient higher hydrocarbons were separated under such conditions.
This is due to the phase behavior of such gas compositions. Fig. 3 shows a phase-envelope diagram of a gas mixture from the Rough field containing a high proportion of injected gas.
At the processing conditions, it is seen that cooling the gas from Points 1 to 2 results in the formation of only a marginal amount of liquid with little change to the dew point of the gas. Further cooling to Point 3 results in no liquid formation and no change in the dew point.
Only if the processing pressure is reduced from Points 1 to 4 with cooling to Point 5 would sufficient liquid formation occur to meet the gas-transmission dew point specification. Obviously, this processing option would incur the penalties associated with the need for recompression of the gas.
Of the eight process schemes considered applicable for this scale of operation, the most economic and operationally attractive was refrigeration followed by adsorption. Refrigeration was the main dew point-control process with adsorption as a supporting trim process for achieving the dew points of those gas mixtures containing a high proportion of injected gas.
Two identical process trains were constructed, each consisting of a refrigeration unit and a two-bed adsorption unit. A simplified process flow diagram of one of the processing trains is shown in Fig. 4. The total capacity of the two trains is 1.18 million standard cu m/hr (MMscm/hr; 1 bscfd).
Adsorption has not been used very often for the dew point control of such large flowrates of natural gas. Refrigeration is the commonly accepted process. The main reasons are the lack of relevant published adsorption data and the scale-up for such applications.
After examination of extensive data and experience, however, from the much smaller original Rough field terminal at Easington where the natural-gas dew point was controlled by an adsorption process, it was apparent that adsorption had several advantages compared with alternative processes.
From the cost point of view, refrigeration has high electrical power requirements associated with rotating machinery, while adsorption requires little electrical power. From the operational point of view, quick start-ups and rapid load changes are difficult to achieve with refrigeration.
Adsorption has none of these problems; start-up and load changes can be virtually instantaneous, a key feature especially during response to changes in gas demand.
PILOT-PLANT WORK
It was realized early on in the support work to the Easington terminal design that mathematical models of the hydrocarbon adsorption and regeneration processes would be very powerful tools. They could be used to study the effect of changing variables such as feed-gas flow-rate, composition, and regeneration conditions.
It was also anticipated that the mathematical models could provide the basis for process-design work. The decision by the process contractor to recommend the use of adsorption in the gas-processing trains at Easington gave an added incentive to their development.
Although the models had not been developed in time to affect the design at the Easington terminal, they have subsequently proved to be extremely valuable. They have allowed methods of increasing the processing capacity at Easington to be identified and have been extensively used in design studies for other natural-gas processing plants.
Although adsorption appeared to be an attractive proposition for overcoming the Rough field processing problem, it was necessary to confirm that a suitable commercial adsorbent existed.
Because relevant published adsorption data were not available, a need existed for pilot-plant investigations to identify a suitable adsorbent. These investigations were performed at British Gas' own gas treatment test facility (Fig. 5) which is used to obtain data on many aspects of adsorption in connection with natural-gas processing.
Experience has previously shown that the data from the pilot plant agree well with that from a full-scale plant. This allows the pilot-plant data to be applied with confidence to commercial applications.
Pilot-plant investigations are carried out over a wide range of operating conditions and most natural-gas compositions can be simulated. Hence it was possible to evaluate adsorbents for the removal of higher hydrocarbons from the range of gas mixtures likely to be produced from the Rough field.
Three potential commercial adsorbents were evaluated: molecular sieve, activated carbon, and silica gel.
The molecular sieve was rejected because its adsorption capacity for higher hydrocarbons was lower than that of the other two adsorbents. The adsorption performance of silica gel and activated carbon were similar and both were potentially suitable.
Comparison of the regeneration characteristics of these two adsorbents, however, highlighted the greater ease of release of hydrocarbons from the silica gel. Consequently, it was chosen as the most suitable adsorbent for the Rough field application.
ADSORPTION MODEL
The mathematical model of hydrocarbon adsorption which is based on operational and pilot-plant data simulates the complex adsorption of hydrocarbons on to silica gel.
For natural gas flowing through a fresh or regenerated bed of silica gel, all hydrocarbons are initially adsorbed to some extent.
The greater adsorption affinity of the higher hydrocarbons can result in the partial displacement of lighter hydrocarbons.
Breakthrough of a component occurs when no sites are available for adsorption. At this point its concentration in the outlet gas starts to increase.
At "breakthrough" (Fig. 6), the concentration of a component in the outlet gas increases up to the inlet concentration, producing the familiar S-shaped (breakthrough to exhaustion) curve.
In natural-gas processing, partial displacement of a component from the bed can result in the outlet concentration of that component exceeding its inlet concentration.
In the breakthrough of natural-gas components (Fig. 6), the alkanes break through in order of increasing molecular weight. The aromatic and cyclic components have their own order lying among the alkanes.
The adsorption Of C1-C4 components is of little importance as they are only adsorbed to a small extent (usually during the cooling stage of regeneration) and they have little effect on the gas dew point.
In the model, a comprehensive data base from extensive operational and pilot-plant tests is used to calculate the quantity of hydrocarbons adsorbed. The outlet-gas composition is calculated throughout the adsorption period and is used to calculate the gas properties (dew point, calorific value, Wobbe number, etc.).
At the end of an adsorption period, the hydrocarbons remaining on the bed are used to estimate the quantity and composition of the condensate produced on regeneration. For a "closed" system such as the Easington design, any hydrocarbons in the regeneration gas that are not separated as condensate are recycled back to the inlet gas (Fig. 4).
Predictions from the model have been compared with data from the Easington absorbers. A comparison of predicted and measured processed gas dew points is shown in Fig. 7.
Good agreement is shown throughout the adsorption period. Confidence in the predictions of the model has enabled modes of absorber operation to be identified which result in lower operating costs by reductions in the frequency of regeneration and therefore savings on fuel-gas usage.
REGENERATION MODEL
The mathematical model of hydrocarbon regeneration has been developed from a heat-transfer model successfully applied to uprating studies on molecular-sieve CO2 purification units at British Gas' LNG sites.'
For any gaseous component attracted to the surface of an adsorbent, an equilibrium exists between the loading on the adsorbent and its concentration in the gas. This equilibrium is affected by temperature and pressure.
Equations to express this equilibrium have been developed, modified to account for dynamic equilibrium, and built into the model.
For given values of the hydrocarbon loading on the bed and the regeneration gas temperature and flowrate, the model calculates the desorption rate and the outlet-gas temperature.
Predictions from this model have also been compared with data from the Easington absorbers. A comparison of measured data and model predictions in terms of inlet and outlet gas temperatures is shown in Fig. 8. It shows good agreement throughout the regeneration period.
The only noticeable difference is in the rates of increase in the outlet-gas temperatures as they approach the maximum value. This results from the assumption made in the model that the multicomponent desorption can be represented by an equivalent single hydrocarbon.
In calculating the total heat required for desorption, the maximum temperature attained by the bed is of crucial importance. Because the gas properties of this single component are representative of the multicomponent mixture, this temperature is unaffected and the model predictions remain valid.
The model has been used to identify where reductions in the regeneration time can be made. These reductions have enabled increased adsorption flowrates to be achieved as a result of reducing the time required for heating. This has also resulted in lower operating costs by savings on fuel gas usage.
MODELING PREDICTIONS
Predictions of the performance have been carried out for both adsorption and regeneration.
The initial adsorption predictions, carried out before commissioning of the Easington terminal, indicated that the adsorption unit alone could process all the gas compositions likely to be produced during the first production period without the refrigeration unit operating.
It was anticipated that this gas would consist mainly of injected gas and thus not contain a large proportion of higher hydrocarbons. These predictions were confirmed during the first production period.
Further adsorption-model predictions for the Easington absorbers (Fig. 9) identified that the adsorption unit could process gas containing the greatest proportion of higher hydrocarbons, i.e., indigenous Rough field gas.
Again, this could be achieved without refrigeration. Indeed, flowrates up to double the design value were predicted to be possible if the regeneration time could be reduced from the design value of 6 hr to slightly less than 3 hr.
Modeling the regeneration of the hydrocarbon saturated beds showed that regeneration in less than 3 hr was possible with the existing processing equipment.
This depended on regeneration being achieved by means of a heat pulse, as opposed to heating all of the bed.
With a heat pulse, only the uppermost part of the bed is initially heated. The remainder of the bed is heated as the flow of cooling gas pushes the heat pulse through the bed.
Combining the modeling predictions of adsorption and regeneration revealed the full processing potential of the adsorption unit. The predictions showed that the adsorption unit could process all gas mixtures from the Rough field, without refrigeration, for flowrates up to double the design.
EASINGTON TRIALS, DEVELOPMENTS
The commissioning of offshore production wells in April 1987 resulted in the onshore terminal at Easington being supplied with indigenous Rough field gas which contains the highest proportion of higher hydrocarbons.
During this period the refrigeration unit was not in operation and the opportunity was taken to obtain further performance data.
This enabled the measured performance of the adsorption unit to be compared with predictions from the models.
The comparisons confirmed the accuracy of the predictions for both the adsorption and the regeneration models.
The trials showed the following:
- The adsorption unit alone could process all potential mixtures of indigenous and injected gas.
- For most processing scenarios, the adsorption time could be extended, thereby reducing the frequency of regeneration and substantially reducing operating costs.
- Regeneration of the absorbers could be achieved within a 3-hr period by means of a heat pulse. This results in a considerable saving in fuel gas and also allows the adsorption unit to process flowrates up to double the design value.
The plant is now being modified to capitalize on the additional processing capacity identified at the Easington terminal,
The two process trains, each consisting of a refrigeration unit and a two-bed adsorption unit in series (Fig. 4), are now being reconfigured as four separate independent processing trains. The combined processing capacity of these four trains will be 3.54 MMscm/hr (3 bcfd), three times the original design processing capacity.
Each refrigeration unit can process 59 Mscm/hr (500 MMscfd) and each adsorption unit 1.18 MMscm/hr (i bscfd).
To take advantage of this additional capacity, further developments have taken place:
The maximum gas-production rate from the Rough field has been increased. Gas from another reservoir, the Amethyst field, is due to be processed at Easington later this year.
ACKNOWLEDGMENTS
This article has been published by permission of British Gas plc. The authors would like to thank colleagues, particularly Mr. R. Wyatt, for their help in its preparation.
REFERENCES
- McHugh, J., "on target with Rough," Gas Engineering and Management, July/August 1985.
- McAllister, J., and Parsons, P.J., "Meeting hydrocarbon dew point requirements," The Chemical Engineer, April 1987.
- Parsons, P. J., Redding, P. S., and Wyatt, R., "Improvements in adsorption processes for treating natural gas," Communication 1275, Institution of Gas Engineers, 51st Autumn Meeting, November 1985.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.
