Modelling piping networks can be a challenging engineering task and sometimes it can be a mixture of both science and art, especially because there are a large number of factors to take into account before starting to build a model. As a process simulation engineer you don’t want to build a model for an incompressible single phase fluid in a straight pipe using a fancy multiphase correlation, or even worse, to build a highly simplified model that will give you an answer so far from the real behaviour that it is useless. I don’t promise that this article will be a panacea for piping network modelling, however I intend to cover the critical steps that you, as a simulation engineer, need to go through in order to successfully model your piping network and get the best possible answer.
Define your Goals!
An essential aspect to cover when modelling piping networks is to clearly define the goals. There are two major scenarios to consider: design and rating. Design corresponds to a case when flow and pressure are known (boundary conditions) and through the application of good engineering practices, the size of every pipe branch is calculated to meet the requirements. Alternatively, during rating, the pipe sizes are known and the ultimate goal is to determine whether or not the system will be capable to deliver the expected flow. This may target network troubleshooting to solve possible deliverability issues (bottlenecks). Finally, when expanding an existing pipe system it may involve a combination of scenarios, i.e. evaluating the existing system (rating) and sizing the expanded section (design).
As with any simulation model, you need to accurately characterize your fluids. Fluids need to be characterized properly for two main reasons: to have an acceptable prediction of those physical properties that affect pressure drop (density, viscosity, among others), and to have appropriate prediction of phase equilibrium due to changes in pressure and temperature. Pure components as well as pseudo components (using Oil Manager) are available in HYSYS for this purpose.
The other important consideration is related to the complexity of the pipe network. Is it a straight pipe? Are there multiple branches? Is the flowrate known? Or do you want to calculate the flow distribution in a network? Is it single phase or multiphase? Is it compressible? All of these questions need to be answered even before opening the simulator. The answers will determine the strategy you should follow.
Keep it simple, use HYSYS Steady State if you can!
In case you want to model a fairly simple pipe network you can always use HYSYS in steady state mode. The evident benefit is that you may be already more familiar with this mode. In addition, you may be able to reuse an already available fluid characterization from an existing HYSYS case. In HYSYS, the Pipe Segment unit operation has already built-in a solver that allows you to calculate the following cases:
- Case 1: Calculate outlet pressure given inlet pressure and flowrate,
- Case 2: Calculate inlet temperature given inlet flowrate and outlet pressure,
- Case 3: Calculate inlet flowrate given inlet and outlet pressure.
Cases 1 and 2 are typically easier to solve compared with Case 3. The latter may be hard to converge especially at relatively high flowrates. It is always useful to verify the pressure data by running the pipe using different flowrates to allow you to bracket the solution.
If you are using multiple pipes (parallel or in series), you can only use HYSYS steady state mode if the flowrate is known. The reason is that HYSYS steady state does not have a pressure/flow solver. If flowrates are known, for example if you have a measuring device in your pipeline, the steady state mode can be useful. There are some ways to trick the simulator to solve the pressure/flow network by using the Adjust unit operation, but we will cover this technique in a future article.
Multiphase correlations are also available in steady state mode. Correlations such as Tulsa Unified Model (2 and 3-phase), Beggs and Brill, HTFS and OLGAS are available in HYSYS. They are useful if you want to evaluate the liquid holdup of a pipe in order to locate a drip-leg device.
Loops, Parallel Piping? Don’t be shy, use Dynamics!
In case you have a piping network that includes branches in parallel and/or in series or even loops, HYSYS Dynamics is a good option because of its pressure/flow solver. Please take into account the following considerations:
• Start your model in dynamics from the beginning: some simulators (including HYSYS) have the capability to go back and forth between steady state and dynamics. However, since both solvers are different, you may crash the model by doing this. My recommendation: start your model in dynamics and never go back!
• Track your specifications: in dynamics you must specify pressure or flow for every Inlet and outlet stream. As a general rule if you have “n” inlet streams and “m” outlet streams in your system, you need to have (n + m) pressure/flow specifications with at least ONE of them being pressure (Streams Dynamics Tab).
• Run your model frequently: it is good practice to run you model for a couple of seconds every time you add a new pipe segment or a valve into your model. Of course, you will need to change your specifications frequently, but by doing this you will make sure that you pipe is always initialized. Running too many pipe segments that have never been initialized may cause your case to crash due to several reasons such as: pressure/flow solver convergence issues, pipes were not properly initialized, or a flash calculation failure.
• Create Stripcharts: plotting pressures and flows as a function of time is the easiest way to verify that your model has reached steady state.
• Model Volume Holdup: if you are NOT interested in the transient behavior of the network, i.e. your ONLY target is to get the steady state “picture”, you can disable the Model Volume Holdup. This will neglect the pipe volume but pressure drop calculations will be still valid. This will help to increase the real time factor particularly for large models (Pipe – Dynamics Tab).
But not everything about dynamics is good; there are two major disadvantages when using HYSYS Dynamics related to piping networks:
• No-Slip Condition: for multiphase fluids this means that liquid and vapour phases are assumed to travel at the same velocity, which is not realistic at relatively low flowrates. For liquids at low velocities, gravity forces are of greater significance compared with inertial forces causing the liquid to travel at lower velocities with respect to the gas phase and producing liquid accumulation (holdup) and potentially slugs. However, if you have a single phase fluid, or reasonable low or high gas-to-liquid ratio (it means close enough to the single phase condition), HYSYS Dynamics will provide good enough accuracy.
• Incompressible flow: in the case of single phase fluids, HYSYS uses the Darcy equation for pressure drop calculations. For gas flow at high velocities, compressibility effects become significant causing a deviation from the Darcy equation. A well-known rule of thumb suggests that if the density (inlet vs. outlet) changes by less than 5%, the gas can be considered an incompressible flow. In those cases, HYSYS dynamics will be a suitable tool. A Gas Pipe unit operation is available in HYSYS, however, please note that it requires an inlet flow specification which is not useful when attempting to model networks.
Complex problems require complex tools!
If your target is to model a piping network with a multiphase fluid (i.e. vapour/liquid or vapour/liquid/water) in steady state or transient analysis, your will need to use HYSYS Hydraulics. To use this functionality the pipeline needs to be configured inside the “Aspen Hydraulics Sub-Flowsheet”. This flowsheet allows you to solve the pressure/flow network along with multiphase non-slip correlations which can be used in steady state or dynamics. If the target is to perform a pigging analysis, this is the right tool to be used. Nevertheless, it is computationally expensive, it means that if you are working with a relatively big network, it may take several minutes to converge a steady state model, or in the case of dynamics, it could result in a real time factor less than 1. For more information, please check the following articles: “Pipeline engineering studies using dynamic simulation” and “Using process simulation for sizing and construction scheduling of gas gathering systems”.
For compressible flow piping networks (e.g. relief networks) our recommendation is to use Aspen Flare System Analyzer (formerly FlareNet), which is a tool specifically developed for flare systems but can be used for general purpose compressible flow network simulations. For more information check the following articles in our database: “Can a general process simulator be used for flare system design and rating?”, “Optimizing Flare system performance – reducing backpressure”, and “Solving highly-looped systems in Flarenet”
Final Remarks – Be one step ahead!
Before diving into building your pipe simulation model, I would recommend you spend a few minutes answering the following questions:
•How is my pipe configuration? Is it a straight pipeline? Do I have parallel branches? Are there any loops in the network?
•What is the fluid? Is it liquid, vapour, mixture? Is it a pure component? Do I have reliable physical properties? Do I expect single phase or multiphase?
•Am I interested in transient behavior?
•In case that the fluid is a gas, are high velocities expected in the network? Is density expected to change more than 5%?
•Am I interested in pigging scenarios, slug analyses?
Answering those questions and using some of the guidelines that I proposed in this article will help you make the appropriate decisions early on to define your modelling approach. The idea is to keep you one step ahead, investing some time at the beginning of the project in order to choose the more appropriate approach. This will increase your productivity and at the same time provide the accuracy you really need.
Do you have questions or comments regarding this article? Click here to contact us.