By Maiki Vlahinos and Ryan O’Hara, nTopology
Using nTopology’s advanced geometry kernel it is now possible to produce a next generation high-performance Heat Exchanger (HEX) for the aerospace industry, as shown in Figure 1, using advanced materials and manufacturing methods. When coupled with ANSYS CFX, the evaluation of high-performance designs can be achieved in ways that were not previously possible.
Figure 1: Triply Periodic Minimal Surface high performance HEX for applications in aerospace turbine engine applications
In aviation, thrust is required to propel air and space-craft through the atmosphere. The engine combusts fuel and extracts mechanical work from this combustion to generate the thrust required for flight. In all engines, the process of combustion and mechanical work produces excess heat that must be dissipated. Specifically, the oil in the engine needs to be cooled to maintain the lubrication of components that rotate within the engine. In modern aircraft, the fuel spends much time stored in the wings, where it gets extremely cold. As such, it can be used to cool many of the subsystems of the aircraft. A FCOC exchanges heat between the engine oil and the fuel in such a manner that the engine oil is cooled whilst the fuel is heated up. This exchange of heat serves two purposes: the cooled oil properly lubricates the engine while heating the fuel prevents the formation of ice crystals within the fuel.
The design shown here was inspired by an America Makes project where it was required to leverage additive manufacturing on a legacy shell and tube HEX for both part replacement and to discover whether advanced design and manufacturing could be used to increase the performance of the legacy component shown in Figure 2.
Figure 2: American Makes Fuel Cooled Oil Cooler in a shell and tube HEX configuration.
Delivering Increased Thermal Performance In Space-Constrained Volume
Many aerospace capabilities are built upon hardware platforms that often cannot be changed without serious modifications. As such, it is imperative that design engineers are enabled to do more with less. One way this can be achieved is by using an advanced geometry representation to mathematically and precisely control the geometry within the interior volume of the design space. In this example, nTop Platform was used to define a volume that could be used to iteratively design a modified FCOC that maximizes surface area while minimizing mass (thickness) of its interior walls. With these constraints these are the only two ways to increase the performance of a HEX.
Heat transfer through a wall can be calculated as:
q = U A dT (1)
And the Heat Transfer Coefficient (HTC) is:
U = k=s (2)
k = thermal conductivity (W ) (3) s = material thickness (m) (4)
Maximizing surface area can be accomplished by utilizing a Triply Periodic Minimal Surface (TPMS); one known as a gyroid, which has both a high strength to weight ratio and very high surface area to mass ratio, is used in this case study [Gyroid = S in(x)Cos(y) + S in(y)Cos(z) + S in(z)Cos(x)]. By using a gyroid structure in the HEX, a 146% increase in surface area was achieved when compared to a more traditional tube-and-shell HEX of the same size. When coupled with advanced manufacturing methods, these TPMS structures enable parts with both high-strength and heat-dissipative requirements to be designed in a manner that was previously impossible to achieve.
To minimize the wall thickness of the HEX, a cutting-edge nano-functionalized high-strength 7000 series aluminum alloy (7A77), that has been developed specifically for additive manufacturing, was chosen for fabrication. Through the increased strength of this alloy the wall thickness of the FCOC was minimized while still meeting critical burst-pressure structural requirements of the aircraft. With nearly twice the yield strength of AlSi10Mg (a traditional cast grade aluminum alloy for AM) the walls of the gyroid can now be approximately half the thickness of previous designs. By using nTop Platform to design the internal core with a gyroid structure it was possible to increase the surface area by 146% and reduce wall thickness by half, which increased the overall heat transfer of the FCOC by approximately 300% within the same volume as the legacy design.
Computational Fluid Dynamic Simulation for Predicting Performance
ANSYS CFX, an advanced computational fluid dynamics solver, was utilized to evaluate the performance of the FCOC. Throughout the design iteration phase several CFD simulations were used to evaluate the design. Driven from initial simulation results it was possible to redirect how the energy was being distributed inside the gyroid, thereby increasing the total heat-transfer coefficient by an additional 12%. A repeatable workflow was developed from nTop Platform into ICEM (for mesh refinement and conversion) and ANSYS CFX, aiding in rapid design iteration.
Figure 3: HTC values with the oil velocity streamlines are shown in the color map on the left with fuel HTC while the color map on the right shows the fuel velocity streamlines with the oil HTC.
Fuel and oil fluid properties and boundary conditions at mass flow rates of approximately 0.45 k g /s and 0.3 kg/s were used respectively. The left image in Figure 3 shows a contour plot of the heat-transfer coefficient inside the fuel domain as well as the streamlines of the oil. The image on the right in Figure 3 depicts a contour plot of the heat-transfer coefficient inside the oil do-main with fuel streamlines moving through the gyroid. With a gyroid core that is only about 100mm (3.9in) in height and 60mm (2.4in) in diameter the overall performance was 3KW (10,200 Btu/Hr).
Now we will focus on the procedural steps that were used to deliver the advanced capability that was previously described. The process for translating geometry from nTop Platform to the chosen CFD tool is summarized by the process shown in Figure 4. The process is defined by the user isolating the fluid domains of the HEX, producing volume meshes of these fluid domains in nTop Platform, and then importing these fluid-volume meshes into the CFD tool, applying the appropriate boundary conditions, and then solving the fluid simulation.
Figure 4: This flow chart depicts the process workflow necessary to get into CFD from nTop Platform. It can be used for a single or multi fluid domain HEX.
The initial design concept for the FCOC went through several iterations on paper as well as in Computer Aided Design (CAD) before entering nTop Platform. The main design considerations were: minimizing pressure drop, enhancing flow characteristics, introducing impingement to improve the HTC, and design for additive manufacturing. As shown in Figure 5, hot oil enters the top pipe (1), moves around the blue dome, enters the gyroid (depicted as a red cylinder), enters the inner diameter and exits out the pipe at the bottom (2). The cold fuel enters through the bottom left opening (3), impinges on the oil outlet pipe, moves up through the gyroid, impinges on the blue dome and exits top right (4).
Figure 5: Original design concept of the Fuel Cooled Oil Cooler (created & depicted in Creo)
The CAD bodies and surfaces seen in Figure 5 were used to define the volume of the HEX and then leveraged to design the infill volume for a TPMS structure. The sketcher and revolve tools in Creo were used to generate the outer shell and dome structures of the HEX.
Heat Exchanger Design Using nTop Platform
After the boundary representations were finalized in Creo, the assembly was saved as individual parasolids and the bodies were imported into nTop Platform. Once imported, in order to properly leverage CAD geometry in nTop Platform, it was necessary to convert the part(s) to an nTop implicit body.
nTop Platform has the unique capability to create TPMS structures in a Cylindrical Coordinate system, Figure 6. This is beneficial to HEX design more broadly and fluid flow in particular. The ability to create these structures in a Cylindrical Coordinate system is also beneficial because you get a symmetric/uniform shape that is circumferentially continuous, as opposed to the gyroid structure that is created in a Cartesian Coordinate system. These gyroid structures are also self-supporting and lend themselves to not only being structurally and thermally efficient, but also readily fabricated in a variety of AM processes without the need for additional support structures during the building process.
Figure 6: Cylindrical Gyroid
Using nTop Platform we can vary the circumference count, radius & height periods, cell size and wall thick-ness of the gyroid to meet the design requirements of the HEX as shown in Figure 6. With that level of control, the designer can tailor the shape of the gyroid to meet performance requirements such as surface area and cross-sectional flow area. This geometric control also allows the designer to adjust the way the fluid will enter and exit to minimize overall pressure drop while optimizing the system-level performance of the HEX. Figures 7, 8 and 10 show how the cell size, circumference count, and height period can be adjusted to achieve a smooth fluid passage throughout the HEX.
Figure 7: Comparison of the fuel exit geometry from the gyroid core
Figure 8: Comparison of the oil exit geometry from the gyroid core
Figure 9: Various inlet configurations were considered to maximize flow and manufacturability of the HEX during the design process.
Figure 10: A comparison of two gyroid designs where the cell size, circumference count, and height period are constant but with the radius period varied. The image on the left encourages the oil to not fully enter the gyroid but rather go straight down the outer shell. Whereas the image on the far right encourages the oil to flow into the gyroid.
Up to this point we have imported and converted our CAD geometry into nTop implicit bodies and generated our fluid domains. The next step in the design process of the HEX will be to create the baffles or flow diverters. These keep the two fluids from mixing. A simple Boolean Intersect block is used to create the baffles. The primary challenge in this step is generating the volumes used to intersect with the fluid volumes. This may require the designer to convert extra CAD entities (faces, edges, vertices) as well as assign parametric control parameters so that as the CAD geometry changes the workflow will be repeatable. Once the intersecting volumes have been generated it is just a matter of selecting the appropriate fluid to block out.
The majority of intersecting volumes were created from extracting CAD surfaces, which were converted to nTop implicit bodies and thickened. The other intersecting volumes used primitive geometry blocks to generate new geometry. The primary block used was the torus, which was then remapped, to create an arched passage way, as shown in Figure 9, that produced a structure that was more amicable to additive manufacturing. Figure 11 depicts the blocks & steps associated with generating the baffles for the FCOC.
Figure 11: This screenshot depicts the nTop Platform blocks that form the gyroid core and fluid volumes. These numbered “parent” blocks require inputs to be complete. These inputs are other blocks and/or general input parameters that control the design and influence the performance. Block (1) Gyroid Core New creates HEX core, block (2) Fuel Fluid New creates one fluid domain and block Oil Fluid New creates the second fluid domain using a Boolean Subtract of Blocks (1) and (2) from the original body used to create the volumes.
Now that the process of creating the baffles is completed, it is necessary to assemble the newly formed HEX core to the components of the HEX. A Boolean Union is used for these operations. During this process, nTop Platform can seamlessly create a fillet between the periodic baffled structure and the “solid” geometry, in this case, the outer shell that was previously drawn in Creo.
Export to ANSYS CFX
At this point in the design, the validation and verification process begins. Finite Element Analysis (FEA) and CFD can be part of the simulation validation and are often used as a precursor to experimental testing. The discretization of nTop Platform implicit geometry for use in a CFD simulation will be described in this section.
As previously described in Figure 4, now that the fluid domains and HEX walls have been generated it is necessary to generate a volume mesh of these regions. Meshing these volumes is achieved through a relatively simple combination of blocks that discretize nTopology’s native implicit-geometry representation into a series of surface triangles and volume tetrahedral elements as shown in Figure 13.
After meshing is complete, the volume meshes can be exported as an ANSYS Fluent mesh (a CFD mesh file type option available from nTop Platform) and imported into ICEM CFD, an ANSYS module used for mesh refinement, conversion, and as a boundary selection tool. Depending on the type of physics being solved, a user would typically choose either CFX or Fluent solvers. For example, Fluent is preferred for high-mach numbers/supersonic flow while CFX is preferred for turbo machinery and other incompressible flow simulations. In order to set up and define any type of computational analysis, the user must apply boundary conditions to select surfaces. These include, but are not limited to, the fluid inlet and outlets faces. Within ICEM we are able to select individual elements. This allows the user to select the surfaces for boundary conditions. Another example of a boundary condition would be a symmetry plane. A very useful reason for selecting faces is to apply boundary layer meshes and perform simple mesh refinement at a localized area.
Figure 12: This is a depiction of the ANSYS Workbench Schematic for the fluid simulation analysis. ICEM CFD and ANSYS CFX were used to perform the final simulation.
Figure 13: The meshing process inside nTop Platform. On the left of the image is the model tree which depicts the blocks used to create and export the mesh. In the center is the mesh of the HEX core and on the top right of the image is the export window with ANSYS Fluent as the format option.
When the boundary faces are defined and the meshes converted, each fluid domain is imported separately into ANSYS CFX. The faces defined are recognized and can easily be assigned to their proper boundary condition. The fuel and oil inlet mass flow rates were set to 0.45 kg/s and 0.3 kg/s respectively with 0 kPa outlet.
Once the nTop Platform-to-CFD-workflow has been set up you can continue to utilize it throughout the design iteration process. Mesh outputs from nTop Platform can be recognized in ICEM as the design updates, which can then be re-imported and the entire CFD work-flow repeated.
The overall feasibility of performing CFD on complex geometry generated in nTop Platform has been demonstrated. nTop Platform allows the user to create complex geometries (TPMS structures, fluid volumes, smooth lattice-solid transitions), while maintaining complete control over the geometric model, and then easily allows the user to export the geometry outside of nTop Platform for validation and verification. The ability to do such complex operations in a single tool while integrating with external CAE tools is unprecedented and can allow for rapid design iterations to be achieved on complex geometry.