CFD Opti­miza­tion Software: 5 Tips for Select­ing the Right One

When we talk about CFD opti­miza­tion software, we actually think of an entirely auto­mated engi­neer­ing process that covers geometry gen­er­a­tion, meshing, sim­u­la­tion, and design post-pro­cess­ing. All of these single steps are auto­mated to create a closed opti­miza­tion loop where, finally, the optimal design candidate(s) emerge from the process, showing optimal results in terms of the pro­duc­t’s flow behavior (e.g. minimum pressure loss, lowest drag, reduced wave resis­tance, flow homo­gene­ity, robust­ness for dif­fer­ent oper­at­ing con­di­tions, etc.).

Nat­u­rally, setting up such a CFD-driven process requires exper­tise in the appli­ca­tion itself but also in various CAE dis­ci­plines, i.e., the engi­neers need to be familiar with the use of dif­fer­ent CFD and opti­miza­tion software packages. Here are 5 quick tips that will help you to select the right CAE solution for your CFD-based opti­miza­tion task:

Tip #1: Focus on Robust­ness of the CFD Opti­miza­tion Software

The involved CAD and sim­u­la­tion software tools need to be robust for automat­ing them. In par­tic­u­lar, the geometry gen­er­a­tion of new design can­di­dates should not fail. Ideally, it should create only feasible designs that do not violate given geometry con­straints. If your geometry model can keep track of this, you will save a lot of time and manual inter­ac­tion. Many tools on the market are not directly suited for auto­mated batch runs (i.e., non-GUI runs without manual inter­ac­tion). Trying to automate them can be a tedious and, in the end, very expen­sive task on your way to a working opti­miza­tion loop.

Tip #2: Demand Flexibility

The opti­miza­tion projects you work on now can be very dif­fer­ent from the ones you will work on in the future. This is not only due to the variety of com­po­nents that can be found in com­pa­nies that need opti­miza­tion, but also due to the degrees of freedom avail­able for explor­ing and dis­cov­er­ing inno­v­a­tive product ideas. When you start doing CFD-driven shape opti­miza­tion, at some stage, you will encounter the wish for more para­me­ters and higher flex­i­bil­ity, while also requir­ing highly effi­cient pro­ce­dures. Hence, your opti­miza­tion loop should be flexible in terms of the appli­ca­tion, no matter if you have a highly spe­cial­ized complex task or a simple opti­miza­tion problem.

For instance, the tools you select should come with com­pre­hen­sive pos­si­bil­i­ties for cus­tomiza­tion, such as script­ing and pro­gram­ming capa­bil­i­ties. From our expe­ri­ence at FRIEND­SHIP SYSTEMS, we can observe that flex­i­bil­ity through script­ing is absolutely essen­tial for creating the most com­pet­i­tive products on the market. So, make sure your software choice con­sid­ers long-term devel­op­ments in your company by pro­vid­ing high flexibility.

Tip #3: Use CFD Opti­miza­tion Software from Experts

Probably, you will first check whether your existing tools can be part of your opti­miza­tion workflow so that no addi­tional exper­tise or software is required. For instance, for a subset of opti­miza­tion tasks, tra­di­tional CAD tools such as Catia, NX or PTC Creo can be auto­mated and used in the opti­miza­tion loop.

However, for more sophis­ti­cated opti­miza­tion tasks, it might make sense to check ded­i­cated software packages, such as CAESES. If you are design­ing complex products, where the automa­tion of geometry gen­er­a­tion and CFD sim­u­la­tion is chal­leng­ing, you should consider software that already solves many of the typical bot­tle­necks, e.g. robust­ness issues during vari­a­tion and details such as keeping the face IDs.

Since you are com­bin­ing several dif­fer­ent CAE tools in a CFD-driven process, your selected tools should also smoothly work together from a tech­ni­cal point of view. Vendors of CFD opti­miza­tion software often have part­ner­ships, to make sure the tools work together without any issues (e.g. data exchange between CAD and meshing tools).

Besides, take a look at the team behind the software that you license: What are their exper­tise and their absolute focus? Can they be a good partner? What about their support, how fast is it? Which com­ple­ment­ing services are offered to accel­er­ate your work on your very specific opti­miza­tion task? Finding a good and spe­cial­ized partner will be impor­tant in the long run, and will create value and benefits for both parties.

Tip #4: Think about Your Geometry Opti­miza­tion Approach

Gen­er­ally, you have a choice between para­met­ric CAD, morphing and defor­ma­tion, topology opti­miza­tion — also to utilize other approaches such as adjoint CFD tech­niques. Let’s take a quick look at these:

Para­met­ric Models

If you want to control your shape through a para­me­ter set that is nat­u­rally con­nected to typical prop­er­ties of your product, you would pick a para­met­ric CAD tool. For instance, turbine or ship hull design­ers typ­i­cally create their geome­tries based on a para­met­ric descrip­tion, that contains items such as profile para­me­ters, chord lengths, contour curves, dis­place­ment func­tions, etc. In these cases, you will create and work with a fully para­met­ric model that provides direct controls for all the prop­er­ties you need. These controls can then be linked to the opti­miza­tion tools or the inte­grated sampling and DoE strategies.

Morphing and Deformation

If the para­me­ter­i­za­tion itself is not impor­tant, you can evaluate morphing and defor­ma­tion tools as your CFD opti­miza­tion software. Changing the geometry is done through defor­ma­tion boxes or radial basis func­tions (RBF), for instance. This allows you to quickly define some degree of geometry vari­a­tion and there­fore, to optimize a product in a shorter period because you do not need to invest the time for creating a para­met­ric model upfront. Instead, you work on imported, i.e., existing or already meshed geome­tries. With morphing tech­niques, you will have limited flex­i­bil­ity and direct control. You will have to somehow ensure that you create only feasible designs and useful geome­tries (e.g not highly squeezed surfaces or meshes, con­sid­er­ing space con­straints, thick­nesses, dis­tances, etc.). Also, if you are morphing meshed geome­tries, you might have to consider how to bring your opti­mized shape back into CAD at the end of the process.

Adjoint CFD

There is also the field of adjoint CFD, which might be of interest if you are dealing with a high number of para­me­ters and rather expen­sive CFD com­pu­ta­tions. If your CFD tool offers an adjoint solver, you will just need one addi­tional cal­cu­la­tion to find out how the dif­fer­ent areas of your geometry influ­ence the chosen objec­tive. These so-called shape sen­si­tiv­i­ties can be directly used for morphing the geometry, or you can connect your para­met­ric CAD model to the adjoint solution. By mapping the shape sen­si­tiv­i­ties to the geometry para­me­ters, you can quickly identify the most influ­en­tial para­me­ters and sig­nif­i­cantly speed up the opti­miza­tion process.

Topology Opti­miza­tion

Finally, there is also a broad field of topology opti­miza­tion, which is suitable for gen­er­at­ing ideas very early in the design process. This approach does not require a para­met­ric model or any defor­ma­tions to be set up. Based on a def­i­n­i­tion of the avail­able space, it takes into account physical con­straints, such as forces that act on a com­po­nent or inlet/​outlet con­di­tions, and auto­mat­i­cally evolves to an optimal geometry shape. Here, a chal­leng­ing task is often to convert the result­ing geometry to a rea­son­able NURBS rep­re­sen­ta­tion for man­u­fac­tur­ing purposes.

Tip #5: Check Value vs. License Costs

There are probably opti­miza­tion tasks that can be carried out manually with suf­fi­cient results and where the added value of a highly opti­mized product is not bal­anc­ing the invest­ments of new CFD opti­miza­tion software.

Com­mer­cial Packages

However, and since you are reading this article, you probably have a design task that needs opti­miza­tion. If your company or depart­ment is cur­rently strug­gling with a budget for com­mer­cial CAE tools, there is one thing you can do: Open Excel and try to cal­cu­late the benefit — in Dollars or Euros, or another com­pet­i­tive advan­tage — when being suc­cess­ful with the tools! Assume a product improve­ment of 3%, 5% or 10% (whatever you want to achieve or is real­is­tic) and write down how your company and your cus­tomers benefit from it.

Also, think about the engi­neer­ing time that you would be able to save by automat­ing the process of creating geometry variants and eval­u­at­ing them. In most cases, a simple cal­cu­la­tion can show you the real value you obtain from effi­cient CAE tools. If you are not sure about the benefit, sit together with the selected software vendors and try to jot it down. The key stake­hold­ers of your company must have a clear idea of the value when invest­ing in software.

Open-Source Software

Finally, if you have a very strong niche or product focus, as well as enough human resources to explore software capa­bil­i­ties on your own, an open-source software might be an option, too. Examples are OpenFOAM, Python packages or the Dakota tool kit, for which CAESES offers an inte­gra­tion in the context of using response surfaces.

List of CFD Opti­miza­tion Software

To wrap up this article, here are a few selected tools, where we know that the CFD opti­miza­tion software packages are well suited for auto­mated processes. Theses solu­tions can be linked to each other, and there are some part­ner­ships between the dif­fer­ent vendors, to ensure a stream­lined data exchange within the tool­chains. For automat­ing the geometry gen­er­a­tion, there are also inte­gra­tions (add-ons or plug-ins) avail­able, to couple CAESES to other plat­forms and envi­ron­ments such as ANSYS Workbench. 

CAD, Morphing, Para­met­ric Adjoints

• CAESES

Meshing

• ANSYS tools
• GridPro
• NUMECA tools
• Point­wise

Opti­miza­tion

• HEEDS
• Mod­e­FRON­TIER
• OptiS­Lang
• Optimus
• Dakota

Sim­u­la­tion

• ANSYS tools
• Autodesk tools
• CONVERGE
• FLOW-3D
• NUMECA tools
• OpenFOAM
• SHIPFLOW
• SIMERICS tools
• STAR-CCM+
• TCFD
• XFlow

More Infor­ma­tion

If you are inter­ested in software for CFD-driven shape opti­miza­tion, check out the CAESES pages. CAESES is an all-in-one solution for CAD, automa­tion and opti­miza­tion in the context of complex product design with CFD.

ANSYS Users

For ANSYS users, there is also a CAESES ACT app to run shape opti­miza­tions directly within the ANSYS Work­bench.

Aero­dy­namic Blog Post

Check out the article ​“Aero­dy­namic Shape Opti­miza­tion: A Prac­ti­cal Guide” which also covers related issues in the field of aerodynamics.