This Model of the Month focuses on a modeling development process using the supplemental TDFUtility program (supplied with all software from ThermoAnalytics), rather than on a specific application of thermal analysis.
The TDFUtility is a set of utilities combined into a single executable. Each utility is organized into tabs. We focus this month on the GEOMETRY tab of the TDFUtility, and demonstrate how it can be used to improve the mesh and model quality of a simple exhaust-shield component model.
We begin our workflow by creating our surfaces in CAD software, in this case Rhino3D software, available for download and free evaluation (www.rhino3d.com). Rhino's meshing tools are primarily set up to generate high-quality rendering meshes, but can also be utilized to generate meshes for thermal analysis. As such, Rhino is a very low-cost means of generating "good" quality base surface meshes. We will use the TDFUtility program to improve and adapt the original Rhino mesh.
We begin by selecting our meshes in Rhino and exporting them in Wavefront OBJ file format. This OBJ file is opened directly into our RadTherm or WinTherm software. Although our software can open native Rhino3D 1.X file formats, the current release of Rhino3D (3.0) cannot save files to the 1.X format. Before we make changes in TDFUtility, we check the conduction in RadTherm using the "Display Free Edges" setting from the View menu. This highlights breaks in mesh conduction in the model geometry. We move to the Geometry Tab>Edit and choose condense. The condense function welds these free edges together. See screen shots below.
After condensing, the file is saved in our native *.TDF file format. We then open the TDFUtilty program from the RadTherm menu in Windows OS. For Linux or UNIX systems, the TDFUtility executable is stored in the "utils" subdirectory under the RadTherm installation directory. After starting TDFUtility, we click on the GEOMETRY tab. This tab can be used to modify existing meshes in *.TDF files:
The workflow when using the GEOMETRY tab is to open an exisiting TDF file as the input file. Then click the "..." button next to Output File to specify a NEW TDF file that will have been modified by the operation of the Quad to Tri. We choose to simply change the "REV" number on the file. Note that we limit the warpage to 1 degree, thus allowing only quads that are within one degree of "perfectly planar" to exist in the model. We can choose to perform this operation on selected parts, or all parts.
Tri to Quad converts pairs of planan Tri elements to Quads. Quad-to-Tri splits warped quad elements into 2 Tri elements. Because the Rhino meshing tool can often create non-planar quads, we will first use the Tri to Quad function to split up those warped quads.
Open New File: Rev-2
We then open the new file in RadTherm and take note of the changed geometry. NOTE: the original "bad" geometry is stored in a new part called "DELETE THIS PART." You can select this part to highlight and view what elements have been removed (more accurately, what elements will be removed when you delete that part). After examining the elements to be removed, make sure that "Delete This Part" is selected and click Delete Part in the Editor tab of RadTherm. Click yes when prompted to delete the geometry also.
Our geometry now appears without the warped quad elements. But we still have a few items to correct—many extra tris are sometimes created at seams. We will use the tri-to-quad function to merge planar tris into quad elements. Quad elements yield faster, more accurate thermal solutions, and so we try to produce a quad-dominant mesh for our analyses.
After deleting the unwanted geometry, we save this file (...rev-2.tdf), return to TDFUtility, and select the Tri to Quad function. The Tri to Quad function will merge coplanar tri elements into quads for a faster solution and better conduction paths.
We now set version 2 as our INPUT file and create a new entry for our output file (...rev-3.tdf). Again selecting all parts, we click on Calculate to perform the operation and store the new mesh in the (...rev-3.tdf) output file.
Returning to RadTherm, we open the output file and again check out "Delete This Part." We see which pairs of Tri elements have been merged; delete this part and save (...rev-3.tdf).
Boundary Conditions
Satisfied that we have planar quads and tri elements only where needed, we set up boundary conditions for the model. We will not go into detail about the model setup, since our focus is on the TDFUtility. However, we summarize the model boundary conditions as follows: the heat source is the formed steel pipe set up as an assigned/interpolated part type. Our first few runs will be steady state, so we interpolate between the fixed values of 500°C at the inlet and 400°C at the outlet.
Our final run will be a transient analysis using curves for the interpolated elements, which in practice could be thermocouple data. We split our fluid tank into 2 parts—the bottom connected to a water node and an air node in the upper zone—representing a partially filled fuel tank. The heat shield is a simple steel heat shield and the floor pan above uses a 3 Layer part type to represent the steel, insulation, and carpet layers. All external surfaces are connected to open air convection at 20°C.
Adaptive Refinement
We run our model and examine the results—they appear as expected, but since our mesh density is rather low, our temperature gradients are not as smooth as desired. We save our model and return once more to the TDFUtility to perform an adaptive refinement of the mesh, based on these initial results.
We set this file (...rev-3.tdf) as our input and specify a new output file (...rev-4.tdf). We select Refine based on Temperature and select all parts. We expand our boundary by 1 element. Since this routine is based on the gradient values in a file, it can be used to refine based on specific time steps of a transient run, or all time steps. In our first case, we have only a single set of steady state results. Thus we select "update" under time steps and select our single set of results. Setting the maximum change to 5 we subdivide high gradient areas of the mesh enough times to predict a maximum change of 5 degrees across any two adjacent elements. We click Calculate to perform the operation and save the results to our output file (...rev-4.tdf).
Final Mesh and Results
We return to RadTherm, open the new file, and examine the Delete This Part part to see where our changes have been made. We can "hide" the bad geometry using the Hide and Show buttons on the quick menu bar. When we are satisfied with the changes, we click Delete Part to remove the old, coarse geometry, leaving us with the refined version.
Since the interpolated part type uses conduction rules to determine its temperature distribution, the elements were renumbered and need to be reassigned. Had we been wiser, we would not have selected the pipe when we performed the refine. We learn by doing.
Our interpolated elements reassigned, we run the model again, generating smoother results.
Transient Thermal Results
With our geometry optimized for the maximum gradients we expect to encounter, we now create curves for our temperature inputs and run the model as a transient analysis over 15 minutes. By refining and optimizing our meshes early in the modeling workflow, our results will be improved and our results can, under some circumstances, be more accurate.
The TDFUtility offers several other useful functions, and we encourage all users of RadTherm, WinTherm, and MuSES to explore its many benefits. If you have questions about the TDFUtility, please contact us.
View More Sample Models
This sample model uses thermal design to improve the aircraft’s human comfort while reducing the power load on its climate control system.
This sample demonstrates an evaluation of AC pull-down using two different source air temperature profiles and the resulting effects on localized passenger comfort. The Berkeley Comfort Model provides localized comfort results for each body segment computed from temperatures predicted by our Human Comfort Module.
The analysis for this structure involved a prediction of solar loading as well as air circulation patterns throughout the structure. ThermoAnalytics used RadTherm software to determine the need for transmissive or non transmissive glass and predicted internal temperatures of the Cathedral on a sunny days at both half and maximum occupancy.
A prediction of transient thermal comfort for humans driving a car through the desert. It provides realistic, localized loading of the vehicle's air nodes.
A human thermal model demonstrates the effects of solar intensity and ambient temperature. They drink a cup of coffee as the Human Comfort Module simulates their thermal response to the environment.
RadTherm can predict Formula 1 racetrack surface temperatures. Simulations based on weather and track temperature can be a powerful ally in an event where a racer can win by mere seconds.
We simulated a coil-in-tube cross flow heat exchanger to demonstrate RadTherm's fluid stream and convection heat transfer modeling capabilities. The heat exchanger surfaces were created in Rhino3D and meshed with ANSA. The mesh was exported into RadTherm.
Air travel passengers are all too familiar with the thermal discomforts of summer travel. Our DC-10 model provides insight to thermal management techniques that could improve the experience.
RadTherm can model the thermal behavior of complex systems. In this case, a refrigerated truck transports cargo in a contolled environment. We simulate the vehicle's thermal behavior and the cargo system's thermal response.
One of the most frequent complaints about motorcycles is the engine's heat plume. The plume hits the rider when they're idling in traffic or after getting off the highway. This analysis simulates this heat plume and predicts the temperature increase that the rider will experience.
This sample model demonstrates RadTherm's unique environmental effects applied to architectural analysis. You can generate transient thermal results by combining CFD results with RadTherm.
Thermal simulation of a brake system. The brake model is loosely based on the front rotors of a 2005 Ford Mustang.
This model simulates simple thermostatic control of two heating systems. It represents a common heated plate system used to cure adhesives or control chemical reaction rates.
Modeling the forest canopy generates accurate temperatures or IR signatures of objects in such environments. Studying the canopy can find passive "cool zones" for urban park development. Through simulation, the ideal size and location of trees can be planned to provide improved thermal comfort.
We model the development process of a simple exhaust-shield component model using the supplemental TDFUtility program (supplied with all software from ThermoAnalytics). The TDFUtility is a set of utilities combined into a single executable. We demonstrate how it can improve mesh and model quality.
Sample unmanned ground vehicle signature analysis featuring a simple hybrid drive engine and exhaust system and sensor tower. The defense industry needs to consider the thermal (infrared) signature of its systems under various operating conditions and environments. MuSES performs a full thermal analysis followed by a radiance solution in specific IR wave bands corresponding to the sensor of interest.
The Human Comfort Module is available under a separate license to simulate the thermal response of a human body in an environment with significant thermal loads.
