Demystifying “Pushbutton” Approaches for CFD & FEA Design, Analysis, Redesign, & Optimization of Turbomachines
Although there is not just one way to design a turbomachine there sure is one way not to do it; blindly.
A misconception that I commonly see when teaching engineers about fundamentals of turbomachines, as well as when leading design workshops, is that some engineers (mostly the younger generations) envision themselves plugging numbers, pushing buttons and getting results immediately without any real brain power behind their actions.
Nowadays, software packages are an integral part of an engineer’s toolkit, but in the same way that a mechanic would not (or should not) use a screwdriver as a hammer, each software has its own applications and ways to use it.
It is common knowledge that CFD analyses are more of a “see you tomorrow” affair than an “I’ll grab a coffee and I’ll be back”.
Although the fairly recent developments in electronics allow for more computing power while being more affordable, it can still take a significant amount of time to run a good CFD case.
One of the main advantages of running CFD is that there is no need to have an actual, manufactured prototype in order to run an experiment. Prototypes have been known to be mainly restricted to companies/individuals that had manufacturing capabilities and quite a lot of money on their hands. However, with recent advancements like 3D printing, this prototyping is not only possible but is also relatively fast (and getting faster everyday with new techniques being developed).
It comes to a point where it is worth evaluating, qualitatively, each method, however different they actually are.
Although CFD is an extremely common practice in modern day engineering and is immensely useful, it tends to sometimes completely replace actual prototyping and this can create some issues… Indeed, CFD is neither an exact science nor it is always “cheap” (some complex problems can easily cost several thousand dollars in computing costs) but either way it sure has its perks. These two arguments are unfortunately largely part of a general misconception of CFD that decision makers and the younger generation of engineers are often victims of. When managers are given the choice between purchasing a software that can supposedly simulate any physical problem (CFD case) and a machine that can physically build components (manufacturing case) the upfront cost strongly leads these decision makers to adopt the first option.
However, CFD does not always suffice. Results of CFD analyses are influenced by numerical and modeling errors, unknown boundary conditions or geometry and more. Refining your mesh is becoming easier and ultimately leads to reduced numerical errors while, at the same time, increasing your calculation time. Modeling errors can come from misuse or inaccuracy of certain models when trying to simulate real, complex physics like turbulence. And so on to the point that different codes and even different engineers can find some minor discrepancies in the final results of the same case.
This means that less experienced engineers tend to over-trust their results, thinking of CFD as the universal answer to every physical problem. To place (smartly) more confidence in CFD results the codes should be calibrated and corrected based on experimental results that do require prototyping at some point unless a product is wrongly put on the market without proper physical testing – which can happen, unfortunately. Comparing both an original and an optimized geometry in CFD is perfectly possible and realistic but as for any solver a baseline should be created. One cannot simply say he has improved the efficiency of a machine by 2% if the original machine was not analyzed beforehand.
Calibration of the CFD models is based on available data from experiments and this data is often very limited compared to the results that CFD can provide. While a physical test would provide values like power as well as some pressures and temperatures in most cases, CFD analyses can go way beyond this by providing parameters distributions, flow recirculation areas, representation of the boundary layer appearing on the surfaces, etc. that allow getting a good understanding of what is happening to the flow within the machine, which is something that definitely cannot be appreciated in most experimental runs. Beside the mentioned disadvantages that 3D printing has, an important one that is shared with CFD is that the time needed to build a geometry strongly depends on its size. However, CFD can deal with the repetition of an element in a row fairly accurately while the entire wheel has to be manufactured to be analyzed. This sort of restrains rapid prototyping to smaller machines, at the moment.
For these reasons and despite all these “warnings”, CFD remains and will remain an essential engineering tool that provides a good comparison of cases rather than a truly accurate representation of the reality we live in. As a conclusion, CFD still continues to evolve with the recent technological developments and should be supplemented with experimental testing instead of substituting it.
Whether it’s to drive you to work, power up your electronic devices, fly you to your holiday destination (extraterrestrial or not), or even set up the perfect lighting for this Valentine’s Day, your daily life requires power production. Although renewable energies are gaining popularity, many people remain unprepared to make the complete switch to these innovative power sources (except Iceland). Making the things we have more “energy efficient” or “green” has become an attractive marketing tool for many of businesses.
The goal of this test case is to find the gas turbine necessary to produce 58 MW of total net power for the conversion of a steam turbine to a combined gas-steam cycle while providing the highest level of cycle thermal efficiency.