Modular System Approach for Modeling and Simulation of Combat Aircraft Survivability Testing

Modeling and simulation background

Modeling and simulation has a long history, and military aircraft modeling and simulation specifically has evolved substantially from primitive tabletop models to sophisticated computerized simulations. Aerodynamics and hydrodynamics largely dominated the evolution of military modeling and simulations until computerized simulations in the 1960s, when—owing to the quick evolution of both computational power and software design—computerized modeling and simulation began answering many aerodynamic questions once reserved for wind tunnels.

Building on the revolution in aerodynamic simulation, countless military modeling simulation disciplines followed suit. Among them was aircraft survivability modeling and simulation that had previously relied on attempting to shoot, hit, and blow up airplanes and mitigate against the resulting damage. Computer simulation of aerial survivability against both air-to-air and ground-to-air guns and cannons was only marginally beneficial for several reasons, among which are the following:

• • The barrage of bullets is too chaotic to predict accurately and repeatedly.

• • The weapons are relatively inaccurate at range.

• • The bullet is unable to change direction.

The path of flight and behavior of missiles, like aircraft, yield to accurate forecasting. Therefore, missiles have a predictable envelope of performance. Predictable and repetitive behavior is far simpler and more accurate for a computer to replicate than erratic spontaneity, driving the success of computer modeling and simulation for military missile engagements (Ball, 2003).

Although not a perfect replica of open-air testing, today, computer modeling and simulation can sufficiently approximate real-world events to provide a reliable assessment of aircraft survivability (Welch & Pywell, 2012). Open-air testing is so complex that, even from the perspective of pure physics and mathematics in a pre-test laboratory environment, it is not always possible to capture every phenomenon affecting the aircraft in flight; fortunately, however, that is not always necessary. The advantage of flight testing is that the experiment takes place in a realistic environment, where even if not all the active elements are scientifically clear, one can nonetheless make conjectures based on the results from the nearly exact environment where the aircraft will operate throughout its life cycle. Take for an anecdotal example if, in a controlled open-air test environment, an SA-5 surface-to-air missile system shoots down a typical cargo aircraft in five out of five different scenarios, it is likely that said cargo aircraft will not survive flying over an unfriendly country uninvited, whether one understands the exact physics of the engagement or not.

With computerized simulation, the same engagement scenario between a surface-to-air missile and an aircraft might be difficult and costly to capture with sufficient fidelity. Whatever physical models that are not programmed into the simulation model are also not reflected in the simulation output. Consequently, all those phenomena that may not require preliminary computations in open-air testing and are simply observable now must be programmed into computer code or accepted as omitted details leading to an imperfect computerized modeling result. Even if someone went through the trouble of understanding every minute natural and manmade phenomenon acting on the aircraft and its surrounding environment, until quantum computing becomes practical, it is unlikely that enough computer power could be harnessed to process the information in a timely fashion (Hassija et al., 2020). Over the decades of computerized modeling and simulation evolvement, the models have gotten significantly better and more precise, quickly (Nance & Sargent, 2002). Even today, however, computerized simulation cannot fully replace open-air testing for that same reason: imperfect approximation (Senneberg, 2021). Nonetheless, modern modeling and simulation can achieve the figurative good-enough results that provide an approximate probability of survivability within a defined margin of error, relying on those physical computations that account for the 90% solution, figuratively speaking. The more questions that modeling and simulation can answer, the less open-air testing is required.

This state of tradeoff is largely where military modeling and simulation has already arrived today and will be heading in the future: more modeling and simulation fine-tuned to answer more questions coupled with less open-air testing at an often prohibitive one million dollars or more per sortie (Welch & Pywell, 2012). The staggering amount of data that modern computers can process—while still short of the power required to compute all phenomena—allows computerized simulations to replicate an open-air test with impressive accuracy, so much so that often it is enough to fly a fraction of the planned test points simply to confirm the conclusions of the simulation. Although the way that researchers have used computers for modeling and simulation has changed over the years, much of the fundamental methodology remains similar. It has always been a handful of algebraic and trigonometric calculations performed at some frequency a certain number of times (Ball, 2003). The more computational power, the more of these calculations are performed with a greater frequency, allowing for arrival at a more accurate answer quicker. This set of equations used has evolved into an overwhelming number of specialized modeling and simulation tools, creating an inefficient and costly Department of Defense (DoD) modeling and simulation enterprise discussed in the DoD Modeling and Simulation Triad section below.

The contribution of this article is a novel modular approach to modernize military aircraft modeling and simulation framework without over-limiting the various military engineering and scientific disciplines, such as aerodynamics, structures, propulsion, heat transfer, fluid mechanics, etc. This digital approach is original and unique in that it would be the only modular one-stop-shop tool for predicting military aircraft survivability through modeling and simulation. This approach should lower costs, relieve propriety issues, and alleviate problems associated with lack of customization and overspecialization. Today’s aircraft survivability modeling and simulation tools fall into one of two general categories:

• • All-inclusive solutions with proprietary, built-in models

• • Independent highly specialized solutions

The approach offered in this article is nonobvious because, unlike the conclusions and suggestions of most research to date on the subject, the typical all-inclusive models with no external modules or plugins created for niche purposes do not offer the most sought-after approach. The modular approach incorporates a core computational software set that includes the mathematical properties shared by most of the military aircraft modeling and simulation tools. The modular system, or black-box approach, builds upon previous effective ideas that were unsuccessfully implemented—namely the DoD modeling and simulation triad (Joint Warfare System [JWARS]/Joint Simulation System [JSIMS]/Joint Modeling and Simulation System [JMASS]), discussed later—attempting to streamline defense modeling and simulation. Finally, this article investigates modern modular solutions that are leaning in this direction, analyzing their strengths and shortfalls.

Computerized modeling and simulation

At the core of these computer simulations are typically just a handful of equations that model aircraft detection, tracking, and engagement. Among these is the common detection and ranging equation called the radar range equation, which defines the effectiveness of a radar by determining the maximum distance at which the radar can detect a given target as well as providing the range to target. Equation 1 is a simplified radar range equation (Ball, 2003)

• • $s$ is the received (echo) power at the antenna,

• • $\frac{P_r}{4\pi R^2}$ is the omnidirectional power density at a distance of R,

• • $G_r$ is the radar antenna gain factor,

• • $\sigma$ is the aircraft’s radar cross section,

• • $\frac{1}{4\pi R^2}$ is the attenuation in power caused by omnidirectional wave propagation over a distance of R, and

• • $A_e$ is the antenna effective aperture.

The complexity of the radar range equation increases sharply when accounting for weather, ground clutter, low observability, aerodynamic properties, detection mitigation techniques, etc. In fact, additional complex equations may be required to calculate less trivial concepts ahead of inputting a particular value in the solver. Nevertheless, the fundamental equation and theory behind radar ranging remains generally the same.

Another common and critical set of equations specific to aircraft survivability is the missile-engagement equations, depicting the missile’s geometry and the trajectory required to hit the target. Much like the radar range equation, these kinematic equations vary with inputs such as complicated environmental factors, missile flight characteristics, on-board and ground-based electronics, etc. Similarly, for simulations of more complex weapon systems, more sophisticated plugins might be required to compute a more accurate input into the solver. Equations 2–10 below are a relatively straightforward set of equations describing the kinematics of missile-engagement geometry (Devan, 2015):

Together, the radar range equation and the missile-engagement equations, in various forms, make up the mathematical core of almost any engagement-level aircraft survivability simulation model like Enhanced Surface-to-Air Missile Simulation (ESAMS) or Radar-Directed Gun System Simulation (RADGUNS) (Wang et al., 2009). Around this mathematical apparatus sit multiple algorithms that compute and output mathematical parameters, such as trajectory, distance, velocity, etc., that are useful in analyzing survivability. These outputs often are converted into a convenient graphical user interface that outputs informative and readily legible images. In all, entire simulations are forged on a handful of critical modifiable equations to expose and fix aircraft survivability shortfalls.

Whereas the core equations of most models are not all that complex, it is in the fine-tuning of the scenario and the players that introduces the complexity. In fact, to achieve an accurate representation of open-air testing or a real engagement, one might use a separate model for each engagement with different missiles, radars, aircraft, etc., to account for characteristics and systems of the various players in a given scenario, such as military aircraft, surface-to-air system, environmental effects, etc. Without getting outside the scope of the article, the reasoning for model complexity and variety, beyond just different environments, is the complexity and variety of different weapon systems. For example, ground-based engagement systems can have a plethora of different detection, acquisition, and engagement radars that span the whole gamut of frequencies, as well as have many different missiles (passive, active, semi-active, different paths of flight, velocity, control surfaces, etc.) with dramatically different aerodynamics and flight characteristics. Individual missile design varies with complex engineering and missile logic systems that impact the trajectory of flight, detonation distance, engine burn rate and time, control and maneuverability, etc. Similarly, different aircraft have drastically different radar cross sections, countermeasures, maneuverability, infrared signature, etc. It is because of this complexity that defense aircraft survivability has adopted niche models for various functions (Ball, 2003). To date, no good solution exists to overcome or simplify that complexity, and this article will not attempt to argue otherwise; however, the core computations that capture the relationship between objects in a simulation (like the ground radar and the aircraft) remain mostly the same, allowing for a universal solution for these core functions (Steinkellner, 2011).

The problem: Too many tools

Despite the relative simplicity and repeatability of these core mathematical computations, countless aircraft survivability modeling and simulation tools exist, including the more commonly known ones such as ESAMS, RADGUNS, Advance Low-Altitude Radar Model (ALARM), BRAWLER, and Trajectory Analysis Program, just to name a few (Hall & Ketcham, 2009). The reason for the existence of so many tools is threefold: proprietary technology, customization, and specialization. Proprietary modeling and simulation tools are often created as part of a major aircraft defense project, an element of a total-package approach to the weapon system in which all sub-components, software, tools, etc., are produced within the scope of the project. Creation of proprietary modeling and simulation tools generally exists for one of two reasons: either the defense contractor creating the tool wants to retain intellectual rights to the software for profit reasons, or a government agency classifies the software at a level open only to a small subset of people. In defense projects with no transfer-of-intellectual-property clause relating to modeling and simulation (and such contracts are many relating to software in general), the defense contractor is not incentivized to create a relatively complex modeling tool with likely no return on investment because the product could be released to the public by the government: such sunk costs run counter to the for-profit business model of most defense-oriented companies (U.S. Government Accountability Office, 2019). By contrast, nonprofit government organizations generally have few reservations about the release of intellectual property, except when such a release might harm national security.

Moreover, these proprietary tools are generally optimized for a specific platform and customized for a certain scenario, making them unsuitable for other applications. Software is usually customized to save costs and improve results with less effort. To limit the scope of the work and thus minimize cost, modeling and simulation tools are customized to fit specific flight characteristics of a platform or a specific environment, ignoring everything outside of the desired flight envelope. Similarly, if a modeling and simulation tool is created specifically for the survivability of a given aircraft with known parameters, the software set can be fine-tuned for specific inputs and outputs, optimizing the efficacy of the model (Hall & Ketcham, 2009).

Finally, and for many of the same reasons, modeling and simulation tools are often specialized for specific combat scenarios like military aircraft engagement with a surface-to-air threat, military aircraft detection by a radar, or some combination of the two, varying in levels of accuracy and precision. Specialization of a modeling and simulation tool creates software that may be good at some computations but less adept at performing others. For example, an engagement model might need an underlying assumption that the target has not been detected upon launch in order for an active or semi-active seeker to have a chance at detecting the target in flight rather than detecting the target before engagement begins. On the other hand, a detection and tracking radar model may not have an engagement mechanism at all. Such a specialized tool may allow for more focus on a smaller, and therefore simpler, problem set. However, this situation results in the need for multiple models to produce a full picture of the combat environment like Figure 1, which is a top-level operational view of a complex combat environment including the various players and assets like aircraft, radar systems, command and control systems, engagement assets, etc. (Erlandsson & Niklasson, 2015).

Figure 1.

 

For example, ESAMS is a long-serving excellent missile-engagement model that simulates “the interaction between an airborne target and a surface-to-air missile (SAM) air defense system” (ESAMS, 2019). ESAMS allows for customization of certain parameters to account for a wide and accurate threat layout and precise aircraft models. However, for example, ESAMS does not account for aerodynamics, so one must use a different model to calculate and output accurate flight data. BlueMax7 was created with that purpose in mind, outputting aerodynamic data files that are compatible as inputs for other models like ESAMS. BlueMax7 produces an accurate flight path with detailed flight dynamics and aeronautic performance. However, although BlueMax7 provides some level of manual interoperability between models, it falls short of being a truly integrated component of a large autonomous or modular system (BlueMax7, 2019). This example repeats itself endlessly in the defense modeling and simulation world, where more and more models are required to develop a full survivability analysis for a given aircraft.

The problem with having so many modeling and simulation tools is that considerable time and money are spent unnecessarily to answer the requisite survivability questions. Usually, many man-hours and multiple expensive tools are necessary to create a full picture of aircraft survivability due to the need for many simulations in many different tools. Moreover, often the tool that best answers a specific survivability question is incompatible with the system under test, or vice versa. In such cases, the solution is typically to create a new custom tool to answer the necessary survivability well-understood questions for the proper platform, exacerbating the need for more man-hours and financial burden. In contrast, a one-stop-shop modular survivability tool with the necessary niche plugins would allow for leaner separate models to answer the same questions, thereby lowering the complexity of tools and man-hours required and minimizing the financial expense of the effort. Efficiency is achieved by eliminating the need for creating the operations common in many modules—the common denominator equations—inside the plugins themselves, capitalizing on the solver. Instead, fewer man-hours and funds are spent strictly on said niche plugin rather than creating an all-inclusive new model, requiring less operations, fewer lines of code, and a more manageable software size (Albers et al., 2019).

In all, there are far too many, oftentimes overlapping, tools to model different aircraft and survivability engagement scenarios, all based on the same fundamental set of equations. Proprietary technology, customized tools, and specialized simulations get in the way of creating one modeling and simulation platform upon which all scenarios can be built. Guarding of intellectual property and declassification of sensitive models creates a great deal of overlap of effort and modeling capability. Similarly, customized model and simulation tools are too narrowly focused, making them difficult to use with an unintended platform. Finally, specialized modeling and simulation tools created for a narrow problem or scenario, without modification for a modular system, similarly prevent the development of a one-stop shop for the entire aircraft combat survivability environment.

DoD modeling and simulation triad (JWARS/JSIMS/JMASS)

The most relevant, but different, past example of a failed approach was the JWARS, JSIMS, and JMASS triad of DoD modeling and simulation, three modeling and simulation environments forming a campaign, theater, and engagement/engineering triad, respectively. The initiative was an attempt to bridge the gap between campaign, theater, and engineering-level modeling while at the same time incorporating many different modeling solutions into a complete package. At the time in the late 1990s and early 2000s, the idea seemed brilliant, only to be overcome by extreme complexity and overstretching of resources—the effort bit off more than it could chew. The modular approach presented in this article focuses on an engineering-enabled engagement-level modular solver only, without overstretching the idea to multiple simulation levels and leaving the customization of models to modular plugins rather than enforcing composability (Maxwell, 2000).

JWARS was meant to be a cutting-edge campaign-level environment that provided battle space situational awareness to decision makers by aiding to digest the complexities of modern joint warfare. The modeling environment included realistic complexities like weather, terrain, logistical limitations, available intelligence, perception of the enemy, and more (Maxwell, 2000). Similarly, JSIMS was a model that accounted for much of the same environmental and external factors but focused more on the operational and even tactical level (Bennington, 1995). Together, these two models were meant to create a complete picture of the entire military campaign down to the individual unit level, with JWARS providing a simpler but broader view while JSIMS dove deeper on a smaller scope or subset of the campaign.

JMASS was meant to take this modeling process to an even deeper level by creating an engineering-level engagement simulation environment with mathematically rigorous probabilities of success. Rather than serving as a model itself, the JMASS environment fell into the composability trap by compiling various existing leading engineering and engagement models that were customized for specific units and tasks (Russell & McQuay, 1993). After simulating an engagement, JMASS would pass a success or failure result back up to JSIMS to display at the theater level view. Conceptually, the idea of an engineering-to-campaign level one-stop-shop modeling environment is undeniably attractive. To this day, the defense community is yearning for such a universal, standardized solution to modeling and simulation for everything from laboratory developmental research to operational testing and the validation, verification, and accreditation process.

The triad undertaking did not take hold in the defense modeling and simulation community, however, because it did not account for the overwhelming complexity of the problem and the variety of models required for different tasks and for different users. The breadth of content and detail required at the campaign and theater levels can be complex to replicate in a computer system, especially when considering the diversity of different operations, units, and dynamics between them. Arguably, though, the true difficulty arises when trying to calculate individual subunit-level engagements. Be they large-scale battles, few-vs-few, or even in the simplest one-vs-one case, once physics and relatively realistic mathematical computations take the place of pre-determined look-up tables, the complexity level skyrockets. Once such complex and processing power-intensive computations are compounded by the number of engagements required at a theater—not to mention campaign—level, the problem set becomes simply too complex. Moreover, given that it is nearly impossible to include every defense solver in one package, the triad came pre-loaded with an unwieldy set of solvers optimized for specific computations while being less than ideal for others, creating a limited composability scenario that was imperfect for most (Davis and Anderson, 2004). Consequently, the JWARS, JSIMS, and JMASS triad, while not inoperable, did not gain widespread popularity in the defense modeling and simulation community, proving to be a jack-of-all-simulation-trades without being a master at any one level. While the DoD modeling and simulation triad is the best example to date of an attempted modular one-stop-shop, a variety of other modular modeling and simulations tools have been created over the years. Each of them either fell shorter in their success compared to the triad or tackled a smaller piece of the puzzle like design modeling, aerodynamic modeling, etc. (Hall & Ketcham, 2009).

The modular system approach presented in this article serves as both a smaller scope and more flexible solution while sharing some similarities with the JMASS engagement and engineering-level component of the triad of defense modeling and simulation. Unlike JMASS, the modular approach would focus on the mathematical and physical backbone required for modeling and simulation computations without creating an all-inclusive limiting package. Various specialized models could be plugged in, in the form of software plugins or modules, into the foundational software set to customize a simulation based on a specific scenario. Similarly, unlike the triad approach, the modular system would focus on the engagement and engineering levels only (about a third of the problem set), minimizing the breadth of complexity. However, although outside the scope of this effort, after customizing the simulation, a user could plug it into some larger-scale theater model or environment, like JSIMS, for example.

In sum, while similarities exist between the unsuccessful JWARS, JSIMS, and JMASS triad of defense modeling and simulation and the modular system approach, JMASS limits the scope and allows for modularity to build on the triad effort’s shortfalls. The scope is limited by omitting the campaign- and theater-level simulation environments altogether (JWARS and JSIMS). Although modularity is introduced by replacing the mostly proprietary and not customizable JMASS simulation environment with a set of building blocks into which models could be plugged in, one can customize the simulation to a unique task. By doing so, the modular approach avoids the insurmountable complexity of multi-level simulations and embraces customization in a narrower simulation scope rather than forcing an all-inclusive solution on the defense community.

Joint simulation environment

The closest existing concept to the modular approach proposed in this article is the joint simulation environment (JSE)—a “scalable, expandable, high-fidelity government-owned, non-proprietary modeling and simulation environment to conduct testing on fifth-plus generation aircraft and systems accreditable for test as a supplement to open-air testing” (Casem, 2019). The JSE effort is a difficult task, trying to incorporate verification, validation, and accreditation, as well as an extensive scope of open-air testing. A key test of most fifth-generation aircraft accreditation efforts is the survivability component. The JSE analyzes the survivability of an aircraft by incorporating an operational flight-program-piloted cockpit with dedicated solver for detection, tracking, and engagement. The environment inputs include “things like weather, terrain, multiple other platforms and air and ground threats” (Casem, 2019), among others, in a seemingly unlimited configuration. Similarly, the brain of the JSE, the solver, is able to output an almost overwhelming amount of data in real time, simulating what may resemble a modern video game, and allowing for a realistic combat simulation of aircraft survivability (Casem, 2019).

Unfortunately, the JSE is not ideal because it fuses multiple different models into one integrated package rather than using a new independent portable solver, partially repeating the missteps of JMASS. Unlike the proposed solution in this article, the JSE utilizes several industry standard software packages for different parts of the total survivability question (i.e., EEAGLES, NGTS, DIADS, etc.) (Smith, 2018). The downside of that approach is the continued reliance on third-party developers for their tools, keeping the costs and proprietary dependability high while limiting customization. Moreover, the JSE is designed from the ground up as a human- and hardware-in-the-loop system, limiting the mobility of the system. In contrast, the approach presented in this article proposes the use of in-house developed solvers and takes modularity to a whole new level, allowing for hardware and human integration and standalone operation (Olivine, 2017).