Modeling of Gas Processing Facilities Using Simscape
Christian Burgstaller, RAG Austria
RAG Austria is one of the largest energy storage companies in Europe focusing on the sustainable use of depleted natural gas reservoirs for underground gas storage and the conversion of renewable energy to hydrogen. Upon extracting the gas from the reservoirs, a main gas-processing step is the dehydration of the gas by extracting water vapor from the water saturated gas stream. Two different dehydration methods are applied: gas dehydration by adsorption to silica gel and gas dehydration by absorption, also known as glycol dehydration.
In this session, we’ll give an overview of the successful development and application of simulation models for both dehydration methods using Simscape®. The physical description of the adsorption and absorption processes could be implemented successfully by applying the Simscape custom component functionalities to build detailed models of the adsorption and absorption column dynamics. We’ll present several use cases that show an excellent agreement with measurement data recorded at the gas dehydration plants.
Published: 21 Nov 2021
Good morning, everyone. My name is Christian Burgstaller. And I'm working as a technical advisor in RAG's underground gas storage development department.
In my talk today, I would like to give you an overview of our applications of Simscape for modeling gas dehydration processes. In particular, I will present Simscape models for cars dehydration by adsorption and gas dehydration by absorption, also known as glycol dehydration.
But first, let me give you a quick overview of RAG Austria, the company I'm working for. RAG was founded in 1935. So we have quite a long experience in oil and gas exploration and production operations.
During the last two decades, our biggest gas reservoirs have been converted to underground gas storages. And we have become a gas-and-energy storage company. We are one of the largest underground gas storage operators in Europe. And our focus is the sustainable use of natural gas reservoirs for underground gas storage and the conversion of renewable energy to green gas and hydrogen.
So the first application I would like to present is modeling of adsorption dehydration units for Simscape. Our initial motivation for this project was to evaluate if Simscape can be used for building dynamic models of cars dehydration facilities with the final goal to build digital twins.
Now, the purpose of adsorption dehydration is to extract water vapor from the gas by adsorption on silica gel. And here, you'll see a picture of one of our underground gas storage plants where we use this method. Now the way adsorption dehydration works is basically that water vapor gets absorbed on silica gel when gas flows through the adsorber from the top to the bottom.
After a certain time, the flow direction is reversed. And the absorbed water gets removed from the silica gel by hot gas flowing from the bottom to the top. And this procedure is repeated in the cyclic manner.
Now Simscape already has quite a big library of components from different physical domains, which is indicated here. But to model the adoption process, there was no such component. So we have used so-called custom components in Simscape to define the physical description of the adsorption process.
In the Simscape model, which is shown here on the right, the adsorber is represented by this yellow component. On this slide, I have summarized the mathematical model that has been used to set up the custom component for adsorption dehydration in Simscape.
The mathematical model is based on conservation laws and the thermodynamics of adsorption. It uses mass conservation, energy conservation, and the linear driving force model to describe the time dependence of the absorbed water concentration. It also uses an experimentally validated model for the adsorption isotherms.
Of course, also the thermodynamic properties of the gas mixture, like specific enthalpy and so on, which are temperature and pressure dependent, have to be included in the model. For this reason, a MATLAB interface to the rev prop database has been created, which imports those thermodynamic properties into the Simscape model.
A typical composition of our gas is shown in this table. The input parameters used in the Simscape adsorption model, the adsorber geometry, the density, and the diameter of the silica gel particles, the ferocity and tortuosity of the fixed bed and parameters that determine the adsorption isotopes. This plot shows a typical adsorption isoterm, which basically gives the equilibrium concentration of the adsorbed water as a function of the water vapor pressure at a specific temperature.
At higher temperature, this equilibrium concentration becomes smaller. And for this reason, the adsorbed water gets desorbed. And the silica gel is heated up.
Here, you see the adsorbed schematics are consisting of four adsorbers for the hydration train. As mentioned before with the adsorption dehydration method, there is a cyclic switching from water adsorption to water desorption, also called regeneration, which is following a certain switching logic on each adsorber.
This switching logic is also based on temperature sensors at specific positions of the adsorber. Now to realize this switching logic in the Simscape model, we have programmed a stateful chart using the MATLAB stateful toolbox.
Now, on this slide, I'm showing some important results of the adsorption model. On the top left, you can see the amount of absorbed water on each adsorber as a function of time. When an adsorber is switched from adsorption to desorption, heated gas flows through the adsorber. And you can see the big temperature steps during that time at the bottom of the plot.
For comparison with measured temperature data, I am showing the plot of the corresponding temperature sensors from our SCADA system. As you can see, we got a very good agreement between the model and real data. What is also impressive is that the Simscape solver can handle the big temperature changes of about 250 degrees Celsius without any numerical problems when switching from adsorption to heating and back to cooling.
Now for validation of the Simscape model, I have done a comparison with real data from January 2019. At that time, we had a feed gas rate of about 90,000 cubic meters per hour. It did not solve the pressure of 90 bar. On the plot on the left, you can see the amount of adsorbed water at different positions along the adsorber as a function of time.
In this case, the adsorber is divided into 10 segments where segment one corresponds to the adsorber inlet and segment 10 represents the outlet. As you can see, the concentration of the absorbed water gets smaller along the adsorber. This makes sense as there is less and less water in the gas that can be absorbed as the gas stream gets closer to the outlet.
Now, the most important result for validation of the model is the simulated partial pressure of the water vapor at the adsorber outlet because this tells us how efficiently the water adsorption is working and if we meet our two point specification. In this example, the calculated partial pressure of water vapor is reduced from 1,130 Pascal at the adsorber inlet to 15 Pascal at the outlet. And this reduction corresponds to a water viewpoint of minus 38 degrees Celsius, which is in very good agreement with the measured dew point data.
I have also used the adsorption model to evaluate alternative operating scenarios. In this example, I have assumed a very high feed gas rate of 310,000 cubic meters per hour, but adsorption train. On the plot showing the amount of absorbed water, you can see that the silica gel in the adsorber would reach saturation over almost the entire length of the adsorber.
On the plot showing the water vapor, partial pressure, we see that the partial pressure at the outlet goes up to 1,600 Pascal towards the end of the adsorption phase. And this would correspond to a water tube point of plus 14 degrees Celsius, which is, of course, outside our dew point specification. So in this scenario, the dew point specification would not be met. And the adsorption cycle time would have to be reduced.
At the beginning of my presentation, I have mentioned that we also deal with the conversion of renewable energy into hydrogen. So gas containing some hydrogen is also of big interest for our future operations. We are actually operating a hydrogen storage research facility where we generate hydrogen by water electrolysis and where we add this hydrogen to the injected gas.
Like in the previous examples, we use an adsorption dehydration unit to dry produced gas. So it was also interesting to see if the Simscape model can correctly handle the hydrogen content in the produced gas. For comparison and model validation of this model, I have, again, compared the simulated water vapor pressure and calculated viewpoint at the adsorber outlet to the measured dew point. Also, in this case, there is a very good agreement with the measured dew point data.
So in summary, I see the following applications for the adsorption dehydration model. It can be used for capacity estimation in the engineering of adsorption dehydration plants. People responsible for operations could use this model to evaluate the energy consumption at various operating modes.
The model could also be used to estimate what the breakthrough times and to optimize the cycle times of the adsorbers. The determination of the achievable dew point and to evaluate the impact of a specific hydrogen content in the gas are further applications.
Now, apart from modeling adsorption dehydration units, I have also used Simscape to set up a model for glycol dehydration units. Glycol dehydration is the second common method for gas hydration. With this method, the water vapor in the gas stream is absorbed in the absorption column by a circulating glycol stream that flows in the opposite direction to the gas.
In the regenerator, the glycolysis heated to about 200 degrees Celsius, and thereby the absorbed water gets removed from the glycol and is vented the atmosphere. Here you see a picture of one of our underground gas storage facilities where we used glycol dehydration. Since there was no ready component in the Simscape library, I have, again, used the custom component functionality of Simscape to build a model for glycol dehydration.
The mathematical model is, again, based on conservation laws, like mass and energy conservation. And in addition, I have applied the thermodynamics of phase equilibria of water vapor and glycol. Similar to the adsorption model, the thermodynamic properties of the gas mixture are required in the Simscape model. And these properties are again obtained from the REFPROP database. The input parameters in the absorption model are the absorber geometry, feed gas, and the glycol circulation rates and the reboiler, temperature, and pressure.
On this slide, you see the Simscape model for a complete underground gas storage facility where we use several glycol dehydration units. The entire facility consists of a central station and two remote stations. The three absorption columns at the central station are represented by these yellow components in the center of the model.
Here we see some more details of the simulation model. Important process parameters are the amount of absorbed water in the absorber, the glycol concentration at the absorber outlet, and the glycol concentration at the reboiler outlet. Also, the gas rates at different positions of the plant can be displayed.
For the validation of the Simscape glycol dehydration model, I have again done a comparison looking at the simulated partial pressure of water vapor at the absorber outlet. The two plots show the water vapor concentration and the water vapor partial pressure along the absorber column. According to the simulation, the water vapor partial pressure gets reduced to 149 Pascal along the absorber.
This corresponds to a water dew point of minus 16 degrees Celsius at the absorber outlet, which is, again, in good agreement with the measured dew point. Calculated glycol concentration at the absorber outlet is 97.7 weight percent, which also makes sense. Also for the glycol dehydration model, it is interesting to see the effects of a certain hydrogen content in the produced gas.
In this example, I have assumed a hydrogen content of 20 more percent but having the same pressure and temperature conditions as in the previous example. This case, the total gas rate would increase from 500,000 to about 570,000 cubic meters per hour, which actually makes sense because gas with this hydrogen content is a much lower gas density. Because of the higher gas rate, more water would be absorbed per hour and therefore, the glycol concentration at the absorber outlet would decrease to 97.3% in this scenario.
Now, what are the possible applications of the glycol dehydration model? Again, I see an application in the engineering of new glycol dehydration units. The operations people could use this model to evaluate the impacts of the feed gas rate, the glycol circulation rate, and the glycol concentration on the water dew point.
Also, the impact of the reboiler pressure and the reboiler temperature on the achievable glycol concentration at the regeneration outlet is a very useful simulation result. And the model could also be used to evaluate the energy consumption at different operating conditions and to estimate the impact of hydrogen in the gas on the dehydration capacity.
So this brings me to my summary and conclusions. As mentioned before, Simscape has an extensive library of components from different physical domains to build complex models. If such components are not available yet, Simscape also provides the option to set up new components using the custom component functionality.
From what I have seen, it has a powerful solver that can handle big changes in input parameters during the simulation. Required input parameters can also be fed into a Simscape model from external sources, like a SCADA system to build real digital trends.
So I think Simscape is a valuable tool for planning and optimization of gas processing facilities. And I am looking forward to new applications in hydrogen technology and renewables. So thank you for following my presentation. And if you have any questions, please feel free to contact me.