Companies utilize different principles to design new turbomachinery. A design exercise is an extremely complex task and requires knowledge of many design trade-offs. This article is intended to reveal preliminary design philosophy and clarify some mysteries in this fast solution method.
Let’s define a few terms first. Boundary conditions (BCs) are the inlet and outlet states of a working fluid. Design inputs are small number of variables that are necessary to begin the design exercise. SoftInWay identifies BCs, design mass flow rate, rotational speed, and a few dimensions as the design inputs. The Preliminary design is a tool for quickly assessing design outputs giving many sets of design inputs. The algorithm utilized in the Preliminary design tool is an inverse solver. Inverse solution in this context implies finding geometry of interest knowing a very few design inputs.
How stuff works? The whole process comes down to estimating losses in each component and then calculating fluid states and component geometry applying simple kinematics and conservation equations. Calculated geometry and states are used to find real losses from loss models. This loss model results are compared with the guessed values and the algorithm repeats until they agree. In a practical implementation, however, the solution scheme will be more comprehensive but underlying principle remains the same — design output heavily relies on the models.
Loss models are extremely important and they determine the range of applicability for an industrial code. The models are collective work of many scientists and designers. Usually, they are some empirical correlations serving large family of components and predicting real machine performance quite well. Can we trust the results? That raises a lot of concerns and skepticism. The predictions are as good as the models that describe the physical processes. Verification and validation plays vital role in the developing of the code. The industry trend is to rely on published scientific data as a first iteration and calibrate models while working on real projects. Range of applicability is determined for each empirical correlation. For example, the veteran of compressor design Ronald Aungier shows that his loss model with respect to return channel in centrifugal stage has good agreement with experiment (Figure 1). Therefore, Aungier’s model can be used for similar machines.
Figure 1 — Loss in optimized return system design
Preliminary design space study — know your limits! When an aerodynamicist is given specification on a new piece of machinery, he/she does not know anything about all the details of the design. Preliminary design can quickly show achievable performance for the machine, estimate critical relationship between design inputs and outputs, and facilitate in determining trends and trade-offs. Design space is a set of many preliminary designs. Because inverse solver is fast, a designer can generate thousands of designs in the matter of eye blink. Moreover, set of mathematical statements and state-of-the art aerodynamic reasoning allows outputting three dimensional geometry for each preliminary design with properly sized components. Ultimately, exploring the design space will eliminate costly mistakes prior to detailed design is carried on.
Myths and misconceptions about preliminary design. Inverse solver does not solve potential flow problem. Inverse task does not perform boundary layer analysis. Preliminary design is not a Navier-Stoks solver. Inverse design is not a table look-up but utilizes empirical loss model in the tested and verified domain. At the same time, preliminary design is not a blade-to-blade analysis tool. Preliminary design is a good starting point for further detailed design and analysis including blade profiling, performance map generation, impeller design, structural analysis, and CFD. All the above can be accomplished within one integrated design environment such AxSTREAM.
Good luck with your challenge!
- Aungier R. Centrifugal Compressors. The strategy for aerodynamic design and analysis. ASME Press. New York. 2000