As the global population moves into cities, gas distribution networks are becoming more crowded. There may be cases when some of the pipelines reach critical saturation of clients and a single event such as an immediate pressure decrease wave can start a chain reaction. Pressure can fall below critical even for a short time, and that can cause problems in the whole grid.
In order to mitigate a risk of such a situation, users can model these events and prepare the grid for emergencies, for example by a small house gas storage. Pipeline simulators traditionally do not model such high speed dynamic cases. We have some real measured flow, pressure and temperature data with a sampling frequency of 20 Hz. We compared the real measured data with our modeled pressure undershoot wave.
We would like to show that the simulated data match the measured data to a high degree. This match especially depends on space and time discretization, when we tried time steps down to 0.1 millisecond. We have encountered some problems, such as very high demands on the calculation time and results database size and some oscillations due to numerical problems of very small numbers. We would also like to show a way how to minimise these demands.
The article presented will also contain tips and tricks and recommended simplifications, in order to be able to simulate very dynamic events and help gas distribution companies model transients in their network with linepack that is next to none.
In conclusion, we present a way how to adjust a pipeline simulator in order to be able to calculate events on a millisecond scale, so that there is an exact simulation of pressure decrease in case of an overcrowded distribution network. We will show that a more universal software can perform the desired calculation to a satisfying degree of precision, so that a specialized CFD package may not be needed.
Gas distribution networks are becoming complicated as the population in global cities is on the rise and many countries are implementing gasification for most of their city dwellers. This increase in complexity raises some specific challenges for simulation of gas distribution networks.
Haulis, Marko (SIMONE Research Group)
One of the necessities of a simulation software is to model the behavior of compressor stations. The main device modeled in a compressor station is a compressor.
The article presented concerns mathematical description of centrifugal and axial compressors, the most frequently used equipment in the gas transportation pipelines or pipeline networks.
Traditionally, both in the machinery industry and in the gas industry, these are mainly modelled using complete biquadratic polynomial approximation of the working space. The characteristics parameterized by the rotation speed values are then represented as 2nd-order parabolas in the satisfactorily invariant (relative to inlet gas status) coordinate system Qvol - Had. Exactly the same approach is basically used for the working space in efficiency variables ?- Qvol. and, parameterized by the efficiency values, also projected (and displayed) in the same invariant coordinate system Qvol - Had.
However, there have been new developments in design and production area, and also the community of the user wishes - namely we are talking about axial compressors that contain several compression stages integrated internally. The characteristics of such devices can no longer be described precisely enough in the traditional simple way represented by 2nd-order parabolas, so better methods of modelling needed to be found.
After some testing of various methods, we propose to model these compressors using linear interpolation approach, when the boundaries of a revolution working area are described using cubic splines and the working point parameters are found using linear interpolation within a small enough quadrangle defined by these splines. There are some challenges using this approach, namely a need to harmonize mathematically precise solution and possibilities of computers. Finally, we realized that the new approach is a bit slower to traditional one, but on the other hand, when comparing the precision, the result is significantly better in favor of the new model.In conclusion, with the advancements of technology, new ways of modelling need to be constantly devised. Our new way of modelling compressors is fast and robust enough to be rolled out to our customers in order to improve the model quality specifically for the compressors that cannot be processed in a traditional way.