RapidIR - Rapid Infrared Signature Prediction

General Description
RapidIR is a technique to rapidly predict the distribution of temperatures across a thermal model. Once the temperature distribution is known, radiance and ray-trace algorithms can quickly generate thermal and infrared images of the target. Compared with conventional thermal analysis codes, RapidIR can predict the thermal distribution 10 to 20 times faster with little reduction in accuracy.

The following thermal images show the agreement between a high fidelity MuSES prediction and the output from a prototype RapidIR simulation. The interaction between the M2 Bradley Fighting Vehicle and the environment, dominated by the rising sun, can be clearly seen in the results from both simulations.

Click picture to enlarge.

High Fidelity                                    Rapid IR

High Fidelity                                    Rapid IR

High Fidelity                                    Rapid IR

Animated Results: M2 Top- Results (652k) M2 Side- Results (406k)

How accurate is the RapidIR technique?
The above images, produced by the first prototype of the RapidIR code, show good qualitative agreement. Future generations of the RapidIR model will correct the small modeling errors seen in the images, such as in the stowage boxes behind the turret ring. Even with this first prototype model, the quantitative agreement is very good. The following four plots compare the temperature of individual elements (i.e., facets) on the turret, a wheel, and the hull. These four areas provide an indication of relative behavior of the entire vehicle. The blue line represents the results from the high fidelity MuSES model, while the black line is the Rapid IR temperature profile.

Wheel Element	Turret Element
Hull - Top Element	Hull - Side Element

In most cases, RapidIR agrees with 1° C with the high fidelity results. At most, the error with this first prototype is less than 5° C. We expect that the maximum error will drop to one or two degrees C in future iterations on the RapidIR model.

How does RapidIR Work?
RapidIR is not a database of pre-calculated thermal results. Nor does RapidIR operate by imposing different thermal distributions upon one another. (For example, a "sunny day texture" is not superimposed over a "Michigan cloudy summer day texture" to create a "Michigan sunny summer day" prediction.) These techniques suffer from limitations.

RapidIR involves the real-time thermal analysis of a carefully crafted low-resolution thermal model. The MuSES thermal solver analyzes the low-resolution model subject to the user-input vehicle operating, environmental, and weather profiles. In the next step of the RapidIR process, the code multiples the resulting vector of low-resolution temperatures by an influence coefficient matrix to produce a high-resolution temperature vector. The code then applies this temperature distribution to the high-resolution geometry.

The two key elements to RapidIR are the low-resolution model and the influence coefficient matrix. The low-resolution model must be carefully created from the high-resolution model to capture accurately the principal thermal phenomena. The influence coefficient matrix, which relates the thermal distributions on the low-resolution model to those on the high-resolution model, must also be generated with care. To create this matrix, both the low- and high-resolution thermal models are run through a lengthy series of simulations. The influence coefficients are calculated such that they account for the variation in facet thickness, thermal properties, mass, and surface optical properties between the elements of the high-resolution model.

The images below illustrate the RapidIR technique in action. The influence coefficient matrix takes the temperatures predicted on the low-resolution model (on the left) and realistically and intelligently "spreads" them across the high-resolution facets (on the right). Note the smooth and realistic temperature gradients produced on the high-resolution geometry.

The general process is summarized as follows:

•   Apply impulse of heat to one high res facet

• Tabulate temperature rise of high res facets grouped into low res facets

• Invert this process to calculate high res facet temperatures given temperatures of low res facets

• [High Res T] = [Influence Coefficient Matrix]X[Low Res T]

The result is more than a low-resolution solution. Previously we showed plots comparing the temperatures of elements. Instead of individual elements, we now look at the average temperature on regions of the vehicle. We find a much better agreement between the RapidIR technique (yellow line) and the high fidelity solution (black line). We also discover that RapidIR improves upon the low-resolution model predictions (purple line). The knowledge of the thermal construction of the target that is embedded in the influence coefficient matrix improves the nature of the thermal solution.

What are the applications for RapidIR?

What are the applications for RapidIR?
With further development of RapidIR, the technique can be applied to a wide variety of applications including:

• Real-time battlefield simulations
• Vehicle design
• Thermal scene modeling
• Design of thermal efficient buildings
• Insulation parameter study
• Operator training simulations

The RapidIR can be combined with other techniques to complement them and improve their effectiveness. One example is instancing. To create a thermal image of a forest, animators construct an entire forest by creating multiple instances of a single tree. The instancing process can be set up to render each tree differently, thus creating the illusion of every tree in the forest being unique. RapidIR can improve upon this technique in two ways. First, the single-tree calculation can be accelerated by using a simplified tree model with simulated results that can then be applied to a high-resolution geometry or texture. Secondly, by varying the influence coefficient matrix, RapidIR can become part of, and enhance, the instancing technique. The end result is that by combining the instancing and RapidIR, one can create a fast-calculating thermal model of a forest that responds realistically to changing weather and solar conditions.

The ability to create influence coefficient matrices that link meshes of differing resolution together, accounting for all the variation in the thermal mass and properties, is useful in applications other than RapidIR. Connecting a computational fluid dynamic (CFD) code to a thermal analysis code would greatly benefit from this approach. CFD and thermal analysis codes require very different meshes for efficient and accurate operation. Using the influence coefficient matrix, boundary conditions (convection coefficients, film and surface temperatures) can be accurately transferred between the two different meshes.

Any design process or simulation that requires thermal analysis under a variety of conditions can benefit from RapidIR. The RapidIR technique can rapidly predict the influences of varying weather, climate, operating condition, and environment (i.e., radiation exchange with surroundings) on any modeled target. A RapidIR-based simulation can predict thermal and infrared scenes that realistically respond to the real-time user input of weather and operating conditions.

IR Signature Modeling Links

MuSES: Infrared Signature Software (Multi-Service Electro-optic Signature) is the next-generation infrared signature prediction program from ThermoAnalytics.  The IR Signature post processor displays physical temperatures, in-band radiances, and apparent temperatures. The user can investigate conduction heat rates, convection heat rates, and radiation heat rates using the post-processor interface. MuSES can project the temperature and radiance data onto the geometry for rapid graphical analysis. The post processor can create spatial or time plots of selected elements and facets. The user can also elect to export the results in the form of tabulated data.

Infrared Signature Model Image Gallery (Examples of models run in MuSES)