Alexey Timoshkin, Deputy IT Director, explains how the Truboprovod team leverages the potential of the C3D geometric kernel to optimize engineering analysis and enhance user loyalty.
NTP Truboprovod is one of Russia’s leading software development companies specialized in the design of oil refineries, petrochemical plants, and chemical facilities. The company has been using the C3D Modeler geometric kernel for many years. Its team regularly attends C3Days, a conference for engineering software developers.
Not all software products from NTP Truboprovod are based on the C3D geometric kernel. The kernel is mostly used in the computer-aided design and engineering analysis tools. The company is developing a range of structural analysis systems.
START-PROF, a pipeline structural analysis system, is the most popular and well-known product from NTP Truboprovod. It is not based on the C3D kernel because pipeline analysis requires virtually no 3D modeling. Only visualization is needed.
PASSAT is a vessel structural analysis solution. This is closer to mechanical engineering, and we use the capabilities of the C3D kernel for modeling.
FEM-FITTING is a finite element tool for strength/stiffness analysis of vessels and pipeline components. It analyzes the structural strength of tie-in assemblies, which are often custom-made. The finite element analysis is largely based on the C3D functionality.
We will talk about these two products and how the Truboprovod team uses the C3D geometric kernel.
Another software product estimates the residual life of the pipeline. The Truboprovod portfolio also includes hydraulic, thermal, hydraulic, thermophysical, and phase equilibrium analysis, etc. Most of these systems support 3D visualization of the analyzed component.
Truboprovod purchased a license to the C3D kernel almost immediately after C3D Labs was established. We released the PASSAT vessel structural analysis tool in 2014, and back then it was based on the 3D kernel. Starting from 2016, we made attempts to integrate C3D in FEM-FITTING. In 2018, we released its version with the C3D kernel. Initially, we applied the C3D kernel for more complicated operations not available in our solutions such as Boolean operations for solids. That was our primary motivation at the time. Once we acquired the kernel, it became clear that we could utilize many of its other features and functions. We found that the kernel offered the features we needed at the time and had already implemented them in-house.
For example, we had proprietary algorithms for calculating model mass and moments of inertia. Since 2019, PASSAT has been using the C3D functions for these calculations. In the same year, we began to use C3D in FEM-FITTING to visualize the analysis models for their configuration. The users now have a live 3D model constructed in C3D instead of static images. These are the milestones of our C3D adoption.
Fig. 1. PASSAT solution
The image shows a vessel with many components visualized in PASSAT. Initially, such models were intended for visualization only: it was more convenient for users to view their design intent. Note that PASSAT primarily adheres to design codes that include only analytical equations. Consider the shell. Its properties are wall thickness, diameter, and length. If we substitute these values into the equation, we get a result that does not need to be visualized. This also applies to the analysis of other individual components such as end plates, fittings, or supports.
Fig. 2. PASSAT solution
The analysis report is like the one shown above. It mimics a manual calculation according to the applicable code. Users appreciate such reports. They can also be used as official documents for certification bodies.
To improve the look and feel, we added approximate 3D models of the equipment. The users reasonably assumed that these models (simplified in comparison to actual CAD models) represented their specific pieces of equipment. But that was wrong. What they saw was just a simplified analysis model that could differ greatly from the actual item. Unfortunately, we had difficulty convincing the users. Focusing on their needs, we gradually came to the point where the analysis software program began to turn into a 3D modeler. It was not planned but turned out to be inevitable. Even symbolically rendered bolts and other small components delight the users. When we analyze vessel supports, the support length is irrelevant. Only the location of attachment points matters. Nevertheless, for better visual appearance these supports should have some specific length, or maybe even different lengths, if the unit is installed on a complex foundation. Our software supports all these functions. While they are unrelated to the underlying structural analysis, they enhance the overall appeal.
Fig. 3. PASSAT solution
This image represents a function that used to be manual and is now performed by the C3D kernel. The vessel is 30% full. The fluid is not only inside the vessel but also in the heat exchanger jacket. We used to calculate the fluid volume with in-house algorithms. Now we get this information directly and seamlessly through geometric modeling with the C3D kernel.
Fig. 4. PASSAT solution: from computational modelling to 3D
What is a so-called analysis element in PASSAT? It is a multi-property analysis model compliant with some code or standard. All the properties shown in the image, or at least 90% of them, are indeed required for the analysis as they are arguments of the equations. Some properties make the visualization look neater. When users input so many values, they expect the resulting model to correctly represent the actual piece of equipment.
Another example is an elliptical end plate. It must satisfy some boundary conditions such as the diameter-to-height ratio, etc. If this is not the case, we cannot apply the standard analytic equations deducted for “true” elliptical end plates. A PASSAT user can specify any shape of the end plate, but if it does not meet the requirements, the analysis is not run, and a warning is displayed. Still, we can visualize such incorrect models. Analysts do realize that complicated types of analysis cannot be done in one click. More efforts may be needed. For instance, it may be necessary to replace the end plate geometry to make sure it complies with the code. To summarize: what we show on screen is not the actual component, but its abstract representation.
Fig. 5. PASSAT solution: from computational modelling to 3D
Below are the visualization results.
Fig. 6. PASSAT solution: from computational modelling to 3D
The model on the left is a standard representation of a horizontal vessel. Such models have been available starting from the early versions of the software. We subsequently added more standard components shown in the middle image: platforms, ladders, etc. These components are considered during the analysis as they apply and receive loads. These components are simple, so no 3D modeler is required. But we need it for more complicated tasks. Suppose we want to cut a hole in the vessel and add a nozzle. This requires a Boolean operation. That is where we began using the C3D kernel functionality which fully met our expectations.
Fig. 7. PASSAT solution: from computational modelling to 3D
The following image shows several analysis models of the horizontal vessel’s support. In real life, there are more. All of them are modeled, analyzed, and visualized.
Fig. 8. PASSAT solution: pipe bundle
Boolean operations for constructing heat exchanger tube sheets. As you see, many tubes pass through these sheets. To accurately calculate the mass, we need to “cut” all these holes with a Boolean operation. It used to take a lot of time. Our team discussed the issue with C3D Labs.
Fig. 9. PASSAT solution: heat recovery unit
Fig. 10. PASSAT solution: heat recovery unit
We managed to accelerate this operation, even when the heat exchanger contained many tube sheets.
Fig. 11. PASSAT solution: customised equipment
Besides using C3D Modeler in PASSAT, we make extensive use of C3D Converter. With it, we export 3D models to other formats and import custom equipment models from other CAD formats supported by Converter. It is a significant benefit. We used to just prompt the user to input the mass. Now the mass is automatically imported, and the model is visualized nicely.
PASSAT uses only the above-mentioned C3D kernel components. The in-house code performs other procedures, including visualization.
Fig. 12. FEM-FITTING solution
FEM-FITTING supports a range of analysis models. We build them programmatically instead of importing them. We define the geometry, mesh it, and run a FEM analysis.
Fig. 13. FEM-FITTING solution
FEM-FITTING supports:
- planar finite elements
- any number of faces sharing an edge
- polygonal meshes
Explanation: Since the software implements quadrilateral planar finite elements, one edge may join more than two faces. That is not the case in 3D modeling. There are certain requirements for the quadrilateral mesh. For example, the mesh angles should be as straight as possible.
Fig. 14. FEM-FITTING solution and geometry
For these reasons, we limit the use of the C3D kernel functionality in some cases. The image shows such a mesh. There is no through hole in the shell. The square tube piece is just welded on. Here, one edge joins three faces. For such a geometry, we rely on proprietary algorithms only.
Fig. 15. FEM-FITTING solution and geometry
Fig. 16. FEM-FITTING solution and geometry
These models look simple but may be challenging to define.
FEM-FITTING uses the C3D geometry kernel functionality: feature intersections and equal radius fillets.
Fig. 17. FEM-FITTING solution
Let’s see how the C3D kernel works to visualize the model shown. It turned out that C3D easily builds all such analysis model configurations. It is sufficient to create one parametric model in C3D, and then edit it.
Fig. 18. FEM-FITTING solution
Fig. 19. FEM-FITTING solution
There are many such models: structural elements, nozzles, with loadcases and impacts. Each of them is built on a C3D parametric model and visualized by Truboprovod’s proprietary code. You can rotate the model for better viewing.
Fig. 20. FEM-FITTING solution
Fig. 21. FEM-FITTING solution
Fig. 22. FEM-FITTING solution
Fig. 23. FEM-FITTING solution
The images show various configurations of such models.
Let us summarize. What C3D geometry kernel functionality do we use? Curves: in many cases, but not always, since some procedures are easier to implement in-house. We also project curves onto surfaces and visualize Boolean operations, meshes, etc.
If the model doesn’t meet the requirements, we run C3D Converter to open the model in Viewer and check it out.
Fig. 24. FEM-FITTING solution
Even such a seemingly standard model can result in twisted surfaces if you change its parameters just a little bit.
Fig. 25. FEM-FITTING solution
The first time I encountered something like this was in a solution then implemented solid Boolean operations using B-splines. The model shown above has no splines, but it still fails in some narrow range of parameters. In a perfect world, the user should have no such problems. This also affects the model's accuracy and applicability to subsequent analysis. Our collaboration continues and we expect to resolve this issue soon.
Alexei Timoshkin
Deputy IT Director
NTP Truboprovod
