A 3D inverse design code applicable to all types of turbomachinery
- Computes the 3D blade geometry from a set of basic inputs and blade loading distribution
- Specifying the blade loading distribution, rather than a blade angle distribution to try and achieve a loading distribution, means that the user has direct control over the 3D flow field
- Provides full 3D inviscid flow-field data for each design iteration
- Solver takes no more than 20 seconds on a standard desktop PC
- Applicable to all types of turbomachinery - compressible or incompressible, axial, mixed-flow or radial, rotating or stationary components
- User-friendly, easy to learn user interface
- The resulting geometry can be easily and automatically exported in formats compatible with all major CAD / CFD / FEA programs.
Basic features of TURBOdesign1
TURBOdesign1 computes the 3D blade geometry for a given distribution of blade loading and blade thickness. Specifying blade loading gives direct control over the 3D pressure distribution and the spanwise work distribution. It allows for direct control of the 3D flow field and allows designers to explore a large part of the design space efficiently leading to breakthrough designs. It results in significant improvement in performance as well as major reductions in design and development time.
Inverse Design versus Conventional Design
There are two methods for aerodynamic design. In the direct method, conventionally used in turbomachinery design, the blade geometry is specified and then modified iteratively by trial and error. Time consuming Computational Fluid Dynamics (CFD) codes are used to evaluate each design and check whether the required flow field, torque and pressure ratio is obtained (see Fig. 1). This approach to a large extent relies on empiricism. In the inverse design method, the blade geometry is computed for a specified distribution of blade loading (equivalent to pressure distribution), which controls the overall behaviour of the flow on the blade (see Fig. 2).
Competitive advantages of TURBOdesign1
Significant manpower and design-time savings
2-3 times improvement in productivity per designer.
Good designs with little prior design experience
TURBOdesign1 does not rely on empiricism and hence it is very easy to train new designers/users who can become productive very quickly.
- Typical improvements of the order of 2-3 percentage points efficiency can be obtained over conventional state-of-the-art designs.
- Significant suction performance improvement in pumps and turbopump inducers.
- Significant improvement in fan noise.
Innovative designs beyond previous experience
TURBOdesign1 allows designers to access a larger part of the design space with ease, thereby allowing breakthrough designs. It has been used for
- Secondary flow suppression and jet/wake flow minimisation in radial impellers.
- Suppression of corner separation in vaned diffusers for pumps and centrifugal compressors.
- Reduction of endwall losses in axial turbine stators without any increase in mid-passage loss.
- Control of separation and secondary flows in torque converter pump and turbine impellers.
- Control of rotating cavitation in turbopump inducers.
Database of design know-how and its transfer
The optimum blade loading for a particular application is quite general and can be applied with ease to similar applications. The results of design optimization studies can be easily implemented in a design database that can be transferred between different design teams in the organisation or to help train new designers.
TURBOdesign1 design process
Step 1: Basic inputs
- Input the initial meridional shape based on 1D sizing code or an existing design. This can be modified within TURBOdesign1
- Input the rpm, blade number and volume flow rate or inlet velocity distribution
- Input the blade thickness distribution
Step 2: Specify blade loading and stacking
- Use built-in calculation window or TURBOdesign Pre output to calculate the average required inlet and exit swirl for the blade row.
- Input the spanwise swirl velocity distribution - both free vortex and non-free vortex distributions can be specified.
- Input the streamwise blade loading distribution based on the main design objectives. This distribution can be fine tuned later.
- Input the stacking location and distribution. TURBOdesign accounts for stacking effects in the blade creation, meaning that stacking becomes an easily accessible parameter with which to control the 3D flow-field.
Step 3: Fine-tuning specified blade loading
- TURBOdesign1 not only computes the blade shape but it also provides an accurate prediction of blade surface velocity, Mach number and static pressure distribution (see Fig. 1).
- Using these parameters the user can very quickly check for problematic flow features sch as excessive diffusion or risk of cavitation and make any necessary changes to loading to fine-tune the design.
Step 4: Off Design Prediction by CFD
- Once the loading is fine tuned, CFD is used to evaluate the off-design performance of the design.
Fig.1 & 2:
What is blade loading and how is it specified?
In TURBOdesign1 the design is based on a distribution of swirl velocity or rVθ. In practice the rVθ distribution is not specified directly but through the specification of ∂rVθ/∂m, which directly relates to the pressure loading on the blade. For example, in incompressible potential flow:
B : blade number
ρ : density
Wmbl :average of meridional velocity across the blade
r : radius
rVθ : circumferentially mean tangential velocity
m : meridional distance
To fully define the blade loading specification for a given design we follow 3 steps:
- Impose rVθ spanwise distribution at the INLET section
- Impose rVθ spanwise distribution at the OUTLET section
- Impose ∂rVθ/∂m distribution on streamwise sections inside the blade
Step 1: Impose rVθ spanwise distribution at the INLET section
Spanwise distribution of rVθ is specified based on exit swirl distribution from the upstream blade row. In the absence of any blade row zero swirl distribution is specified.
Step 2: Impose rVθ spanwise distribution at the OUTLET section
r : local radius
Vθ: tangential velocity
The exit rVθ is directly related to:
- Work input coefficient (rotor):
W=ω(r2Vθ2 - r1Vθ1)
- Spanwise work distribution (rotor)
- Outlet flow conditions (stator)
Step 3: Impose ∂rVθ/∂m (blade loading) inside the blade
How do I choose the optimum blade loading distribution?
The optimum blade loading distribution to control given aspects of the 3D flow field has strong generality across designs. The below publications give an introduction to the typical blade loadings used in some common applications. For information related to your specific application, please contact ADT Sales using the form below below.
- Secondary flow suppression for centrifugal impellers
Secondary flows are reduced by specifiying fore-loading at the shroud and rear-loading at the hub. This type of loading has been found to be applicable to both pumps and centrifugal compressors. Zangeneh et al (1998) (clickable link to paper)
- Suppression of corner separation in vaned diffusers
For 3D diffuser design, users typically specify fore-loading at the hub and aft-loading at the shroud. This type of loading helps to remove corner separation in both pump vaned bowl diffusers and centrifugal compressor vaned diffusers. Goto and Zangeneh (1998) (clickable link to paper)
- Improving pump suction performance whilst maintaining performance
A blade loading distribution in which there is little shroud loading in the first 10% of chord followed by fore-loading results in good suction performance and efficiency. This type of loading has been found to have generality across a wide range of pump sizes and specific speeds. Bonaiuti et al (2010) (clickable link to paper)