Search the Community

Hi, This is a post in the context of project performance: If you use loops with feature definitions, take care where you declare ("create") your participating objects: Most of the times, objects can be declared outside of the loop since the dependency is kept within the loop by means of the expression mechanism! In the attached example, I create N curves in between of two given rail curves and store them in a list for surface creation. The two points on the corresponding rails are declared right before the loop starts. You could also declare them within the loop but - and here comes the message - they would be created again and again, i.e. at each iteration (N-times). In the end, this might slow down your feature execution when it comes to a higher number of objects that get declared within a loop. Hope this helps ... Cheers Joerg loop.fdb

Hi, You may have asked yourself why you may actually need a parametric model? In other words: Why does a conventional CAD model not do? If you are interested in modeling geometry only once, i.e., if you are pretty sure you do not want to change the geometry later, say to adapt to a slightly different design task, well, then it might not be worth the effort to create a parametric model. So, you just produce your geometry in your CAD tool and be done with it. However, if you know that you will have to do variations it should quickly pay off to spend more time up front in order to come up with a model that is somewhat "intelligent." A parametric model is such a thing: You create relationships between various entities that make up your product. Usually, that leads to a certain hierarchy, i.e., some entities become dependent while others are free to change. The latter are the parameters that you want to vary, for instance within a CFD driven design process. They are often called free variables. A very simple dependency may be that some entities are multiples of others, for example B=0.5*L and H=B*L. Often, the relationship of entities is rather complex. Let us look at a slightly more complicated example (not too complicated though): From a plane sheet of paper you can fold a paper ship. Once you have selected your sheet of paper, for example a standard DIN A4, the size of your ship is set. The only free variable left is the beam of your ship. All vertices of your object are now defined by these two parameters, length and beam. If you do not want to tear your paper apart increasing or decreasing your beam will require your vertices to follow, they are dependent. Attached you can find a parametric model of such a ship as realized within CAESES / FRIENDSHIP-Framework. Essentially you get two main advantages from a parametric model: You will always get what you want (since you already put in your design rational you will not create anything that is not really of interest at this point in time)You will reduce the number of things you have to take care of when making changes (and this reduces the degrees of freedom of your system)Let us go back to the paper ship for a second: You could still just define all vertices conventionally in a CAD system. Consider the three vertices shown in the example, V0, V1 and V2. They fully define the shape since all other vertices are just mirrored. Together these vertices offer 3*3=9 coordinates for shape control. Taking advantage of symmetry conditions we can boil things down to 2*2=4 coordinates to be modified; V0 does actually not change at all, V1 can only move in the transversal plane y-z and V2 can only slide in the longitudinal plane x-z. The parametric model intrinsically takes care of this. The knowledge is built in. Consequently, you not only have fewer variables (typically one order of magnitude less) but also no danger of creating something that cannot be produced. In the paper ship model vertex V1 is moved along the circular path in the midship plane. Vertex V2 has to follow such that the distance between V2 and V0 as well as between V2 and V1 is maintained. This defines a circular path, too. Finding the correct corrdinates for V2 is done within an inner optimization (which also illustrates that things can be more complex than just B=0.5*L). Have fun, Stefan parametricPapership.fdb