2 min 5 sec read

In 2014, the Department of Defense (DoD) and the United States Special Operations Command (USSOCOM) placed a heightened emphasis on maintaining a state of readiness among special operations forces (SOF). Given the high demand for SOF that exceeds available capacity, the USSOCOM Operations Directorate (J3) recognized the need for a systematic approach to prioritizing requests for forces from theater special operations commands (TSOCs). In response, J3 developed the Global Special Operations Synchronization (GSOS) model, a comprehensive tool that evaluates and prioritizes SOF and resources globally.

The USSOCOM J35 Division took the lead in developing the GSOS model, collaborating with Operations Research and Systems Analysts (ORSAs) from USSOCOM and the Naval Postgraduate School to ensure an objective and comprehensive approach. The model draws on a range of strategic and operational planning documents, including TSOC Theater Campaign Support Plans, Security Cooperation Plans, and the USSOCOM Campaign Plan-Global Special Operations. As a deterministic, value-based model, GSOS enables senior leaders and planners to make informed, data-driven decisions about force allocation and resource management in support of global special operations.

The GSOS process is a year-long staff endeavor that involves regular consultations and touchpoints among the USSOCOM Senior Leaders, GSOS Council, TSOCs, and other stakeholders. The GSOS model uses algorithms and statistical techniques to analyze inputs and rank campaign activities. The results are visualized in a PowerBI dashboard, enabling senior leaders to make informed decisions about resource allocation and campaign prioritization. The model considers three key inputs: threat weights, campaign activity attributes, and expert voting scores, providing a comprehensive basis for decision-making.

Looking ahead, the GSOS team is focused on enhancing the model's capabilities. This year, the team aims to integrate cost estimates into the GSOS model, providing an objective component to the model. This will allow for a more accurate prediction of deployment costs and inform the planning and programming of the next USSOCOM Program Objective Memorandum. In future years, the ORSA support team plans to build on the model's excursion functionality by introducing a sensitivity analysis tool. The addition of this functionality will help senior leaders identify the most critical factors driving the model's outputs and recommended resource prioritization. Along with the cost data, the team is exploring ways to increase objectivity and reduce subjectivity in the model's human inputs, while acknowledging the ongoing need for human input and judgment in the GSOS process. By continually refining and improving the GSOS model, the USSOCOM can ensure that its SOF remains optimally allocated and effective in supporting global objectives.

1 min 11 sec read

Abstract: This abstract is part of a broader series of abstracts from the U.S. Army Test and Evaluation Command (ATEC), where we will discuss digital transformation efforts underway in ATEC. ATEC’s senior leaders have challenged our workforce to improve the efficiency and efficacy of the test and evaluation portion of the acquisitions process while reducing live testing by 30% by 2030. Various efforts currently underway provide the components necessary to achieve this vision.

The Verification, Validation, and Accreditation (VV&A) process is critical for ensuring the credibility and reliability of Modeling and Simulation (M&S) tools used in decision-making. However, navigating complex and often fragmented VV&A requirements can be challenging, particularly for individuals with limited experience. To address this, the U.S. Army Evaluation Center developed the Accreditation Aid (A2), a comprehensive, user-friendly tool designed to streamline the VV&A process and enhance compliance with key regulations and standards.

Built within Microsoft Excel using Macros, A2 is compatible with any computer for enhanced accessibility and collaboration. This tool is tailored to serve individuals with prior accreditation experience, individuals new to the process, and test officers who frequently accredit instrumentation.

A2 simplifies the accreditation process by providing clear instructions, consolidating essential resources, and offering tools to guide users through self-accreditation. Whether you're a seasoned professional or new to the process, A2 ensures a seamless, efficient, and compliant accreditation experience.

1 min 6 sec read

Abstract: The Army's current modernization efforts are guided by the principle of "No Blood for First Contact". This highlights the importance of leveraging robotic and autonomous systems to drive intelligence, fires, and maneuver to meet the future Army challenges. This integration of robotic and autonomous systems combined with the “expanded battlespace” increases the complexity for battalions and below. With the expanded battlespace, the lowest echelons are not only responsible for the ground and a portion of the air domains, but also saturated air-ground-littoral, cyber, and electronic domains. This requires an integrated, layered network of sensors and effectors that extend engagement ranges and protects friendly formations. This network of sensors and effectors must include levels of robotic and autonomous control that allow Soldiers to control more with less. In the future fight, a single Soldier could be controlling 10-30 systems at once, which is not possible without reliable autonomous control. Maneuver formations equipped with such capabilities will have enhanced situational awareness, and increased ranges for sensing and delivering kinetic and non-kinetic effects at a much larger scale. The capabilities demonstrated at Project Convergence Capstone Five created increased decision space for commanders to employ lethal un-crewed systems to shape the expanded battlefield. This presentation will explore the DOTMLPF-P holistic approach required for the robotic and autonomous transformation that enables expanded maneuvering at echelon.

1 min 6 sec read

The U.S. Army conducts humanitarian assistance and disaster relief (HA/DR) operations across the Indo-Pacific as part of its Theater Army mission to project landpower and sustain regional partnerships. Planning for rapid response in this theater is complicated by vast distances, contested access, dynamic natural disaster patterns, and infrastructure variability. Anticipatory logistics planning—including the siting of prepositioned stocks and alignment of distribution capacity—is critical to meeting operational timelines. This research introduces a two-stage stochastic mixed-integer programming (SMIP) model that supports APS planning under uncertainty. The first stage determines optimal APS locations and stock quantities. The second stage executes scenario-based distribution and vehicle routing in response to simulated disaster events generated via a Monte Carlo engine that varies severity, location, and demand. The model minimizes a weighted additive penalty function over unmet demand and safety stock violations, offering tunable prioritization based on operational risk tolerance and mission objectives. We identify key limitations in deterministic models—particularly geographic biases resulting from disaster-node overlap—and implement Benders decomposition to address scale and improve tractability. Planned extensions include modeling multimodal transportation, commodity heterogeneity, and separating disaster exposure from routing networks to better reflect the operational environment. This framework directly supports U.S. Army Pacific (USARPAC) and the Joint Logistics Enterprise by providing a decision-support tool for scalable, anticipatory planning across contested and disaster-prone regions.

54 sec read

Popular simulation tools for engagement level analysis or higher such as PRISM and AFSIM are single threaded processes. Many analyses using these tools must run many cases with potentially hundreds of replications to understand the average behavior, which can result in millions of simulations for a single analysis. HPC systems offer tens of thousands of CPU cores to facilitate this. However, managing millions of simulations is still an enormous difficulty. Managing data output is also no easy feat, since hundreds of terabytes can be produced in a matter of days. Presented here is a novel piece of software inspired by techniques from the fields of Data Engineering and Supercomputing to manage many millions of simulations and the associated data. This is achieved by leveraging the python package called mpi4py, a SQLite database for metadata, and an API for plugging in a post processing program to collect data from each simulation as it completes (or later if desired). What follows is an explanation of these features, as well the motivations and design choices that went into the product.

3 min 42 sec read

The purpose of this study was to apply topic modeling, a natural language process (NLP) technique, to investigate how individual descriptions of work duties and tasks differ by experience. Previous research has suggested that task ratings may differ by job tenure or experience (e.g., Tross & Maurer, 2000). These data are often collected by gathering information about job duties in job analysis surveys. Surveys can include open-ended items where individuals can further describe specific responsibilities or requirements (Morgeson, Brannick, & Levine, 2019). However, open-ended items often hamper job analysis efforts aimed at understanding contextual information about duties performed have as they can be tedious and time-consuming for human raters to analyze (e.g., Putka et al., 2022). Thus, job analysis efforts often use measures of importance or frequency to describe duties performed in a given position, and do not take into account how duties performed differ by intra-role experience (i.e., time in position or time in rank). This can introduce challenges when Army talent managers attempt to determine what knowledge, skills, and abilities are needed upon hiring or entry into a position versus knowledge and skill that are developed over time. However, advances in machine learning (ML) and NLP have allowed us to quickly and efficiently analyze survey participants' open-ended narratives to determine whether differences in task descriptions emerge by factors such as experience or time within a role (i.e., do Soldiers new to a position report the same duties and responsibilities as a Soldier who might be more seasoned within their work role?). Gaining this understanding can better inform when critically important training or developmental experiences should be provided as Soldiers prepare for particular work roles. These methods can also help researchers and practitioners identify specific contexts that should be considered when assigning individuals (e.g., remote assignments may require a higher level of knowledge or expertise, which may make it difficult for inexperienced Soldiers to perform successfully when they are confronted with new situations where consulting with teammates is not possible).

Data were collected from Army officers ranked 2LT through LTC using an online survey. As part of this survey, we collected data on time in rank and time in position, along with self-described duties. We conducted topic modeling analyses using Latent Dirichlet Allocation on officers' self-reported duties, comparing how topic weights for extracted topics may change when they were new to their role versus more experienced. Initial findings indicate that while many topic weights tended to show similarities across ranks regardless of time in position, subtle nuances were observed in some cases that can be important in Army efforts to modernize talent management, such as identifying talent attributes required upon entry versus those developed over the course of an assignment. For example, LTs in the Military Intelligence Branch tended to show similar average topic weights on topics related to briefing leaders and coordinating unit training, indicating the frequency of these duties tended to be relatively similar regardless of experience. In contrast, Intelligence LTs who had held their rank for more than one year showed increased topic weights for duties related to intelligence collection planning and conducting investigations, which may indicate that as these officers gain experience, they engage in these duties more often.

The results of this study indicate that even when holding similar job titles, some responsibilities and job duties may differ by intra-role experience. In job analytic efforts, this is a critical consideration when analyzing responses from job incumbents that may share job titles, as well as for career mapping and development, succession planning, and considering person-job fit. By considering one's experience and tenure, we also gain insight into how task and duty importance ratings may vary greatly within work units and organizations, along with the knowledge and skills required to perform one's work successfully. Additionally, we demonstrated how using topic modeling to extract position duties and requirements can greatly expedite the job analysis process, as well as use rich data to gain a more comprehensive and nuanced understanding of duties and contexts that affect how Soldiers perform their jobs.

The research described herein was conducted by the U.S. Army Research Institute for the Behavioral and Social Sciences, Department of the Army. The views expressed in this presentation are those of the author(s) and do not reflect the official policy or position of the Department of Defense, U.S. Army, or U.S. Army Research Institute for the Behavioral and Social Sciences.

1 min 13 sec read

Continuous transformation of Army decisions hinges on moving more data through analytic pipelines in less time – without compromising fidelity. Towards that end, DEVCOM Analysis Center (DAC) partnered with the National Security Innovation Network (NSIN) and Texas A&M University (TAMU) to re-engineer and deliver a significant step-change improvement to BRL-CAD's solid geometry ray tracing – the computational backbone for the Advanced Joint Effectiveness Model (AJEM) and many other Army M&S tools. These changes resulted in a dramatic improvement to analysis run times without needing to alter AJEM's physics or code base.

Leveraging computer graphics advancements, refining cache-aware memory layouts, and streamlining data paths with a modern bounding volume hierarchy acceleration structure, ray tracing of triangle mesh geometries was accelerated 2x faster on simple geometries and 20x faster or more on complex models. Changes underwent significant verification, validation, and performance testing improvements in order to accelerate with confidence and to replace two prior triangle shotline interrogation methods. BRL-CAD's permissive Open Source governance facilitated university partnership and allowed for changes to be made directly in BRL-CAD, while AJEM – modularly architected to call BRL-CAD as a service – needed no code changes allowing for seamless integration of the performance improvements without requiring changes to AJEM itself. This presentation details the engineering approach, benchmark results, VV&A methodology, and distills lessons on how to open-architecture M&S ecosystems to better absorb outside innovation and deliver faster analyses. Future opportunities for analytic pipeline improvements are also given consideration.

2 min 16 sec read

The Immersive Simulation (IS) team within the Ground Vehicle Systems Center (GVSC) has been conducting a series of virtual simulations where various ground vehicle components are tested. These simulations are known as the Virtual Experiments (VEs). A VE consists of a large group of active-duty soldiers who are provided with novel protective vehicle capabilities in an effort to evaluate their effectiveness on the battlefield. However, each VE is only able to test a handful of variables and takes months of planning (as well as costs a significant amount) to conduct. This research uses Artificial Neural Networks (ANNs), a type of Machine Learning (ML) method, to provide a predictive algorithm based on past VE datasets. The goal of this research is to accurately predict vehicle survivability based on the VE datasets and thus allow for a greater number of variables to be quickly and more cost-effectively analyzed. While it is common for ANNs to be trained with thousands, if not millions, of data points, the VEs used only has about 150 data points. Thus, a shallow ANN with a few input features is examined with the need to supplement with synthetic data. The ANN setup will consist of the following independent input variables: Enhanced Capabilities Active/Non-active, Vehicle Type, and Mission Type. The ANN dependent output variable consists of one out of two classes related to vehicle survivability: No Significant Vehicle Damage (No Kill) or a Catastrophic Kill (K-Kill). From the available VE datasets, two types of ANNs were analyzed with the following inputs: 1. ANN Type 1: Vehicles A & B - Vehicle Type and Enhanced Capabilities On/Off

2. ANN Type 2: Vehicle A - Mission Type and Enhanced Capabilities On/Off

The ANN architectures comprised of eight variations for each ANN type. One set of ANNs contained a single layer of hidden neurons and a second set with two layers of hidden neurons. The number of hidden neurons was incremented as follows for each layer: 1, 2, 5, and 10. Additionally, synthetic data for both ANN types was generated due to the number of K-Kills being significantly lower than No Kills.

For ANN Type 1, comparing the one hidden layer of five neurons per layer architecture with and without synthetic data yielded significantly different results. The F1 score and Average Accuracy with synthetic data were 0.699 and 76.8%, while without the synthetic data, the F1 score and Average Accuracy were 0.543 and 81.4% (for the test datasets), respectively. For ANN Type 2, comparing the two hidden layer of one neuron per layer architecture with and without synthetic data also yielded significantly different results. The F1 score and Average Accuracy with synthetic data were 0.465 and 50.6%, while without the synthetic data, the F1 score and Average Accuracy were NaN and 88.5% (for the test datasets), respectively.

2 min 25 sec read

The Medical Evacuation Doctrine Course (MEDC) is the Army's only course dedicated to planning medical evacuation (MEDEVAC). My Course Chief, Captain Mahdi Al-Husseini and I wanted to improve medical planning outcomes by providing our students with real-time feedback to their planning. To facilitate this, we started the Medical Evacuation Wargaming Initiative (MEWI). Students will take the knowledge they have gained from the course and apply it to planning MEDEVAC within two Large Scale Combat Operations (LSCO) scenarios. Roles are assigned based on the scenario (Europe or INDOPACOM), and each role is given the ability to place their assets as they see fit and control those assets within a 3D rendered environment. MEWI produces casualties of differing types that require the students to manage the movement of casualties to higher levels of care. The students' goals are to evacuate as many casualties as possible to the Role 3, minimize the died of wounds (DoW) rate, and accurately document the movement of each casualty (medically regulate). In order to achieve these goals, students must move casualties from the casualty collection points (CCP) to the Role 1, send a 9-Line MEDEVAC request to higher in order to have the patients transferred from the Role 1 to Role 2, and again send another 9-Line to request patient movement from the Role 2 to Role 3. MEWI generates a unique patient ID number, and students must use these to accurately track to movement of patients through the Roles of care. The casualty's status is not static within the game, and until they are treated at the Role 2, there is a chance of death. The longer the casualty goes without definitive treatment, the higher the likelihood the casualty will die. Achieving these goals are easier said than done. MEWI produces anywhere from 400-750 casualties per scenario, each scenario lasting around two to three hours real time (12-18 hours in-game). During After Action Reviews, students unanimously stated they were quickly overwhelmed by the number of casualties received. By having every Role 1 overwhelmed, MEDEVAC assets from the Role 2 and air ambulance platoon now had competing requirements with limited assets. The greatest challenge students faced was medically regulating. Students are given the freedom to develop their own regulation procedures and determine the best way to tackle the challenge. Those assigned as medical regulators found that many patients went missing somewhere in the evacuation chain, with no record of their movement until they reached the level of care after the one they were listed as going to next (i.e. patient is at the Role 1 and disappears until they are reported at the Role 3). MEWI provides students with actionable feedback based on their medical plan. Most plans do not last as the original concept but are refined and revised in order to achieve the best outcome possible. Students leave MEWI with a greater understanding of planning medical evacuation in LSCO and the challenges they may face in the future.

1 min 0 sec read

Many organizations looking to adopt generative AI solutions face a critical limitation: existing solutions often cannot securely access private data. To address this challenge, the 1st Multi-Domain Task Force (1MDTF) deployed StormChat, a chatbot powered by Microsoft's Azure OpenAI Service, and designed to provide Soldiers with efficient access to verified, authoritative information. As a new brigade-sized unit with theater-level responsibilities, 1MDTF operates across all five warfighting domains (land, sea, air, space, and cyberspace), which requires rapid access to vast amounts of knowledge for mission execution. Using grounded retrieval-augmented generation (RAG), StormChat ensures Soldiers responses are cited in 1MDTF's curated corpus, eliminating the need to manually search through thousands of files scattered across various SharePoint libraries and minimizing concerns about hallucinated information. This presentation explores key use cases demonstrating how grounded RAG LLMs can be leveraged at multiple echelons throughout the military. We highlight some promising developments, including the incorporation of GraphRAG, Azure AI Vision, and role-based access controls, which enhance StormChat's ability to retrieve context-specific information while ensuring users can only access authorized data. By fostering dialogue on advancements and challenges of LLM integration, we aim to shape the U.S. Army's approach to adopting AI-driven solutions for strategic battlefield advantage.

56 sec read

This mixed-methods study examined how U.S. Army chaplains experience and facilitate the free exercise of religion. Using Brinkerhoff's Success Case Method, the research sent a quantitative survey to 1,478 active-duty chaplains, achieving a 23% response rate, followed by twelve guided interviews representing diverse ranks, genders, and faith traditions. The research revealed six major themes: significant gaps in education and training, the tension between personal religious convictions and professional duties, the crucial impact of leader influence, the importance of proactive support approaches, systemic challenges, and transformation of understanding through direct experience. The analysis employed multiple theoretical frameworks, including Socio-Technical Systems Theory, Open Systems Theory, and Edgar Schein's organizational culture model. Findings indicate that successful religious support requires comprehensive education, psychologically safe environments for dialogue, standardized processes, and intentional leadership development. The study demonstrates how chaplains navigate supporting diverse faiths while maintaining religious integrity, often experiencing significant personal transformation through direct engagement with service members. The research contributes to understanding how religious professionals balance personal convictions with institutional requirements in pluralistic settings and provides the first peer-reviewed application of organizational design theories to the U.S. Army Chaplain Corps.

1 min 2 sec read

Abstract: Long-Range Precision Fires (LRPF) are critical in shaping the operational environment to set conditions for the success of ground forces. As technological advances reduce the time between target detection and engagement, the ability to sustain the volume of fires needed becomes increasingly important. Traditional resupply methods, such as “order-up-to” policies, often result in inefficient or mismatched distribution of munitions, limiting a unit’s capacity to effectively engage targets. This research investigates alternative ordering strategies designed to maximize the number of targets that LRPF units can engage under dynamic battlefield conditions.

A simulation-based framework is used to model target presentation, apply engagement decision algorithms, and implement resupply strategies based on target density and distribution. The model begins with a baseline representing current doctrinal approaches and iteratively modifies both engagement and ordering policies. A linear programming (LP) model, assuming full knowledge of future target sets, is used as an upper bound to benchmark each strategy’s performance.

Findings highlight the trade-offs between different ordering and decision strategies and their effects on maximized target engagement. The results inform the development of a generalized optimization framework that can be extended to other sustainment challenges across the logistics enterprise, enhancing operational flexibility and resilience in contested environments.

1 min 5 sec read

Abstract: Traditional approaches to software reliability often fall short in capturing the complexity and task-driven nature of Army software systems. Failures are typically tied to specific uses for software, rather than occurring randomly over time like seen in hardware systems. This difference drives the inability for traditional methods to identify where the system is most vulnerable or what areas need improvement. Software reliability is typically measured as a single overall value for the system, causing the results to be heavily influenced by the specific tasks performed during testing and making it difficult to draw meaningful conclusions about how the system will perform during a mission.

DAC introduced a new approach to software reliability that shifts the focus from generalized failure rates to a function-centered methodology. By defining a system in terms of its essential functions and decomposing those functions into discrete tasks, the reliability of its essential functions can be assessed using binary outcomes (pass/fail) at the task level. Modeling a mission as a series of these functional uses enables the calculation of a probability of mission success, providing a more practical and intuitive measure of system reliability. This presentation will demonstrate how this approach yields more meaningful reliability estimates and can be used to help prioritize testing efforts by identifying high-impact functional areas.

1 min 12 sec read

The DEVCOM C5ISR Center's Field Repair Information Data Analysis Environment (FRIDAE) application seamlessly integrates various Enterprise Resource Planning (ERP) systems to revolutionize logistics data mining and analysis allowing for data-driven and proactive decision making. As an Army Vantage Workshop application, FRIDAE connects authoritative systems, enabling the extraction and analysis of critical logistics data such as: weapon system material details and relationships, field and sustainment-level maintenance operations, and supply chain actions. By leveraging key data sources within Army Vantage from across the enterprise, the application provides a one stop shop for Army Maintenance Engineers, ORSAs, Logisticians, and Provisioners to meet their leadership's analytical, modeling and simulation needs. The FRIDAE application creates aggregate views of consolidated data from disparate ERPs, supporting multiple use cases. For example transforming basic measurement point data into operating hours for C5ISR systems that can be plagued by OPTEMPO ambiguity. In this example, exposing usage rate data via user-friendly visualizations enables stakeholders to monitor and analyze the performance and utilization of key assets, facilitating informed decisions and optimizing operational efficiency. In other use cases, FRIDAE's sophisticated filtering and multilayer customized text querying supports rapidly-built aggregations. This allows Maintenance Engineers to amass field level Failure Reporting, Analysis, and Corrective Action (FRACAS) data critical to root cause analysis and quality improvement initiatives. Throughout this briefing, we will demonstrate the application's common use cases, its integration process, data sources, and success stories within the C5ISR Center and beyond.

2 min 6 sec read

The Army anticipates a future operating environment in which decision-making cycles will be increasingly accelerated. When decision-making cycles are accelerated, it creates a demand on decision-makers to shift how they are approaching their task. Studies in psychology and behavioral economics have distinguished between System 1 (quick, intuitive) and System 2 (slow, systematic, deliberative) decision-making processes. System 1 processes are emphasized when decision-makers must engage in accelerated decision making. System 1 processes tend to rely on cognitive heuristics, which are mental shortcuts that previously led to, but are not guaranteed to provide, sound decisions. Heuristics present a challenging space for scientific operationalization and assessment as they both increase the efficiency of decision making and introduce a higher likelihood of errors. This presentation focuses on describing a methodological approach to identifying and assessing the types of errors that may be introduced through the application of heuristics in accelerated decision making.

In this presentation, we describe a methodological process by which we (1) constructed a taxonomy of cognitive heuristic errors based on tasks and decisions Army leaders often face, (2) developed an assessment that diagnoses Army leaders' unique heuristic error-making profiles, and our plan to (3) provide tailored interventions to inoculate Army leaders against cognitive heuristic errors due to the demands of accelerated decision making. The proposed presentation will describe the methods employed for construction of the Army cognitive heuristic errors taxonomy, the method for developing the assessment, and describe our progress in establishing evidence of the assessment's reliability and validity. This is a methods-focused presentation, therefore we do not present the taxonomy and measurement tools themselves.

This presentation is proposed for WG 6 – Recruiting, Talent Management, and People Programs. It seeks to improve capabilities for talent management by describing methodology to create tools that will enable the Army to diagnose and measure individual tendencies toward cognitive heuristic errors that may be elicited by the accelerated decision-making cycles anticipated for the future operating environment. This project supports Army leadership doctrine (ADP 6-22) and Army strategic plans (Army People Strategy, 2019; Army Modernization Strategy, 2019), as well as multi-domain operations doctrine (FM 3-0: Operations) and national strategic plans (National Defense Strategy, 2022).

The research described herein was sponsored by the U.S. Army Research Institute for the Behavioral and Social Sciences, Department of the Army (Contract No. W911NF-23-C-0042). The views expressed in this abstract are those of the author(s) and do not reflect the official policy or position of the Department of Defense, U.S. Army, or U.S. Army Research Institute for the Behavioral and Social Sciences.

22 sec read

The Center for Army Analysis's Combat Sample Generator (COSAGE) provides tactical attrition factors used to adjudicate combined-arms ground combat in all Department of Defense theater- and campaign-level combat models. The COSAGE team will present updates to the COSAGE model and modeling methodologies from the last year, including updates to capture the effects of implementing short halts" to engage targets and the overall performance of artillery systems that were assessed using K-shape clustering analysis.

"

1 min 7 sec read

DEVCOM Analysis Center and Sandia National Laboratories (SNL) have developed and demonstrated a repeatable and reliable optical fragment tracking procedure to collect Behind Armor Debris (BAD) in an efficient and cost-effective manner. BAD occurs when a threat perforates an armored target and a cloud of fragmentation, consisting of target and threat material, is ejected behind the armor. Since BAD can inflict damage to internal vehicle components and personnel, understanding of this phenomenon is important to ballistic vulnerability analysis. In conducting analysis to inform system designs and Army acquisition decisions, DEVCOM Analysis Center represents BAD characteristics, such as size, shape, materials, locations, and velocities with probability density functions. Parameters for these functions are derived by matching statistical distributions with results from testing and experiments. The process of capturing and characterizing BAD to improve vulnerability and lethality studies can be tedious, time consuming, and expensive. Traditional BAD experimentation uses a witness pack to capture the debris; manual disassembly and scanning to obtain locations, hole sizes, and penetration depths; then replacing with a new witness pack for each subsequent shot. Advances in high-speed camera technology and finite element (FE) models have enabled DEVCOM Analysis Center and SNL to collect BAD in a more efficient and cost-effective manner. This presentation highlights the optical fragment tracking experimental process, accomplishments, status, and planned path forward for the initiative.

35 sec read

Wargames provide valuable insights for complex military problem-solving but require rigorous data collection to enable effective follow-on analysis and modeling. This presentation details data collection efforts and lessons learned from a recent Center for Army Analysis (CAA) wargame involving participants from numerous Army organizations and other services. As part of this effort, the team developed new data collection templates to capture relevant data during wargame execution. The team subsequently aggregated this data to develop model inputs into the campaign model, JICM (Joint Integrated Contingency Model), such as initial starting conditions, unit structure, and ground combat actions. This work helps provide practical insights for the operations research community to effectively translate wargame results into robust modeling and analysis.

1 min 57 sec read

AEC Digital Evaluation Tool (ADET) for Enhanced Test and Evaluation

This abstract is part of a broader series of abstracts from the U.S. Army Test and Evaluation Command (ATEC), where we will discuss digital transformation efforts underway in ATEC. ATEC's senior leaders have challenged our workforce to improve the efficiency and efficacy of the test and evaluation portion of the acquisitions process while significantly reducing live testing by 2030. Various efforts currently underway provide the components necessary to achieve this vision.

A significant digital transformation effort within the U.S. Army Evaluation Center (AEC), a subordinate of ATEC, is the AEC Digital Evaluation Tool (ADET). ADET is designed to revolutionize the current evaluation process by providing a configuration-controlled digital environment that ensures traceability of measures to their corresponding data within the ATEC Data Mesh. This tool will facilitate evaluation planning, execution, and reporting, thereby enhancing the efficiency and effectiveness of Test and Evaluation (T&E) efforts.

ADET supports digital threads from requirements to data sources, with the long-term goal of offering a digital interface to Army Digital Acquisition Environments. A key component of ADET is the development of reusable measures, which are standardized measures of performance and effectiveness that can be consistently applied across various systems. These reusable measures enable automated connections between measures and data sets from all data sources, ensuring consistent evaluation approaches and rapid development of Data Source Matrices (DSM).

This presentation will provide an overview of ADET, its goals for traceability, and the process for developing reusable measures. We will discuss the steps involved in creating a reusable measures dataset, including identifying key domains, binning similar measures, and consolidating DSM elements. Additionally, we will highlight the benefits of ADET in streamlining data management and analysis, improving data integrity, and facilitating data-driven decision-making within ATEC.

By creating a seamless digital connection from evaluation planning to all leverageable data sources and back for reporting, ADET is envisioned to facilitate AEC's transformation into a data-centric organization. This presentation will be of interest to evaluators, subject matter experts, and stakeholders involved in T&E, providing insights into the future of digital evaluation and the role of ADET in achieving ATEC's digital transformation goals. Additionally, those pursuing or interested in digital transformation may be interested in overarching lessons learned to date in ADET's development process.

1 min 1 sec read

This presentation provides a case study of the data analytics the First Armored Division employed in support of "Ironclad", an initiative to improve Soldier quality of life and reduce harmful behaviors. This analysis leveraged data to identify high-risk populations, understand the root causes of harmful behaviors, and systematically provide resources to at-risk Soldiers. Utilizing data from Army systems of record, surveys, and internally produced historical records, we employed statistical modeling to explore relationships and identify key risk factors. Findings revealed statistically significant, albeit low-to-medium, correlations between high operational tempo, negative unit climate, lower manning levels, and increased rates of harmful behaviors. Additionally, analysis revealed that the true prevalence of harmful behaviors were often multiples higher than reported through traditional channels, highlighting a critical gap in awareness. Existing initiatives like resiliency screenings now have been modified to enable future assessments of effectiveness. As a result, Division Holistic Health and Fitness resources have the information they need to target their approach, and leaders at all levels have a better understanding of the needs of their individual Soldiers. Remaining challenges include improving data integration and evaluating the long-term impact of these interventions. Future study will focus on assessing the program's effectiveness and identifying additional preventative measures.

43 sec read

This project develops a value-based methodology to evaluate and optimize U.S. Army
Pacific (USARPAC) exercises in the Indo-Pacific region, aligning them with Operation
Pathways' strategic goals and resource constraints. In an environment marked by China's
growing military capabilities and heightened regional tensions, U.S. defense leaders have called for more resilient force postures and deeper cooperation with allies.

Through structured interviews with USARPAC personnel and direct involvement in exercise planning, the project builds a value model—a structured framework that quantifies the relative benefit, cost, and strategic impact of different exercises. By analyzing historical and forecasted data, the model supports data-driven decision-making to prioritize exercise attributes that offer the highest return on investment. This approach enables USARPAC to conduct the right training, at the right time, with the right partners, ensuring maximum strategic impact while maintaining cost-effectiveness in a rapidly evolving security landscape.

1 min 15 sec read

Abstract: Effective personnel assignment in airborne operations is critical for aligning load plans with ground tactical objectives while minimizing risk. Existing methods primarily rely on a bottom-up refinement approach, incrementally adjusting load configurations without holistic evaluation. This process often fails to prioritize individual jumpers based on mission-critical attributes and lacks robust mechanisms for assessing load plan feasibility, a limitation that intensifies with larger assault forces. We propose a top-down optimization strategy to quantify jumper value and assess load plan feasibility.

We define mission-critical features for each paratrooper—such as rank, weapons system, team assignment, and vehicle operation status—to quantify their strategic value. Leveraging these values, we introduce a multiple knapsack model where individual aircraft are treated as separate knapsacks with capacity constraints, and Paratroopers, each with an assigned strategic value, are the items to be optimally loaded. Our approach simultaneously balances team composition and rank distribution while maximizing operational effectiveness.

Our method quantifies load plan feasibility and assesses risk through a structured grading system, that evaluates factors like team distribution and equipment/skill distribution across aircraft, offering a decision-support tool for operational planning. This analysis enables planners to quickly identify high-risk load configurations and make adjustments prior to execution, enhancing both safety and mission success.

We also employ random search metaheuristics to generate contingency reallocation plans, ensuring resilience against last-minute manifest changes. Our work introduces a novel, systematic, and optimization-driven methodology for generating resilient and tactically sound load plans in airborne operations planning.

54 sec read

Abstract: The Regionally Aligned Readiness and Modernization Model (ReARMM) is the Army’s current force generation model and became fully operationally capable at the end of 2023. This model is meant to support modernization of Army equipment and formations at a scale not seen in the past 40 years, while simultaneously maintaining enough unit readiness to support global campaigning demands. Grounded in three foundational principles, ReARMM seeks to provide predictability, stability, and synchronization for the Total Army.
This presentation will highlight the flexibility of ReARMM to adjust to emerging needs. Combat aviation brigades (CABs) have significant transportation and maintenance requirements that reduce the time available for these units to modernize and train prior to mission phases. The Center for Army Analysis (CAA) worked with stakeholders to develop novel concepts to reduce friction points in the CAB ReARMM mission lines. These candidate solutions ensured that CABs still met all operational requirements, while providing more dedicated time to modernize and train. CAA then led a multi-objective decision analysis effort to quantify and assess the candidate solutions, leading to a final recommendation to senior decision-makers.

49 sec read

Abstract: Responding to a direct request from the Under Secretary of the Army, articulated in the FY24 Army Analytic Plan, this project developed a novel methodology for aligning science and technology (S&T) investments with the Army Warfighting Concept (AWC) for 2030-2040. This crucial analysis bridged the gap between future operational needs and current S&T planning by assessing the S&T Program Objective Memorandum (POM) against the AWC's Concept Required Capabilities (CRCs). A rapid, collaborative effort delivered actionable insights to Army leadership and contributed to the S&T investment decision-making process. Furthermore, the project facilitated in-depth analyses of critical insufficiencies in S&T development, leveraging workshops with over 675 experts from 50 Army organizations to establish a robust technology baseline. Interactive analytic dashboards visualized key findings, illuminating risks, trade-offs, and opportunities for external partnerships. This data-driven approach, initiated in response to the FY24 Army Analytic Plan, provides a transformative model for future S&T planning, ensuring investments effectively empower the Army of 2040 and beyond.

1 min 10 sec read

Rapid, accurate combat loading of personnel and equipment is crucial for Army sustainment operations and success in large-scale, contested environments. Tools like the Integrated Computerized Deployment System (ICODES) Single Load Planner module assist load planning personnel, but optimizing complex combat priorities often requires time-consuming manual adjustments and significant cognitive effort, limiting overall planning efficiency. This research introduces an automated, prioritized two-dimensional bin packing (2D-BPP) approach to enhance combat load planning, aiming to improve operational effectiveness and logistics readiness. We formulate the problem as a mixed-integer linear program (MILP), minimizing a weighted objective that captures multi-level tactical priorities. The model integrates essential constraints: non-overlap, obstacle avoidance, item separation, and global load balancing (via center-of-gravity limits), which is crucial for transport stability. While smaller instances can be solved directly via this MILP using commercial solvers (e.g., Gurobi), this approach does not scale effectively. Therefore, for large-scale problems, we employ a two-stage framework. Stage 1 uses a sliding-window decomposition method for an initial plan optimized for tactical priorities. Stage 2 applies an iterative post-processing MILP adjustment, performing minimal modifications to meet load balancing constraints while preserving Stage 1 priorities. Results with realistic Army equipment data show this two-stage framework rapidly generates high-quality, balanced, tactically sound load plans, offering significant improvements over current methods. By automating tactically prioritized layouts, this work provides enhanced decision support for Army planners, enabling faster, more adaptable load planning and bolstering sustainment operations in complex operational environments.

1 min 6 sec read

The study "AI in Military E-Learning Mgmt Systems" puts an emphasis on finding ways to improve the individual soldiers learning experience by using AI. The developed prototypes were verified during actual military training at the Engineers School in Ingolstadt and the Army NCO School in Delitzsch to evaluate the actual benefits to the trainee. This approach ensures a fast and robust identification of promising AI applications in E-Learning.

The study is conducted by the Germany Army Headquarters in a joint effort with the Army Training Command in Leipzig. Many use cases for the use of AI in military training have been tested: Learning recommendations based on individual performance, LLM Chatbots, adaptive exams and instructor dashboards with personalized learning recommendations. All with the goal of providing a highly adaptive learning environment tailored to the trainee that can be fine-tuned by the instructor. The different AI applications are integrated into the central E-Learning Mgmt platform to ensure that no compatibility and system integration issues arise. All applications are implemented in an unclassified environment and according to the General Data Protection Regulation.

This presentation will outline the different AI applications developed during the study, present findings regarding the effective application of AI methods in an E-Learning environment and provide guidelines in the responsible and competent use of AI in Military E-Learning.

1 min 9 sec read

In support The Army Talent Management Task Force (ATMTF), a novel methodology has been developed to inform artificial intelligence agents on a bespoke, talent management decision framework (methodology) to optimize the assignment of soldiers to billets using a data-driven, artificial intelligence-enhanced framework. This methodology, piloted under the Knowledge, Skills, and Behaviors (KSB) Innovation initiative, addresses a longstanding gap in dynamically aligning personnel capabilities (supply) with mission requirements (demand). The Marketplace Matching methodology leverages the Army Research Institute developed Army Talent Attribute Framework (ATAF) to quantify individual and job-specific KSBs through semantic similarity analysis across multiple data sources, including Officer Evaluation Reports (OERs), skill assessments, and self-professed attributes. These inputs are synthesized into a unified scoring system that evaluates the "level of match" between soldiers and billets. The system then generates and evaluates all feasible Courses of Action (COAs), both constrained and unconstrained, to recommend optimal assignment slates. Key innovations include the integration of constraint-based optimization (e.g., MACP, EFMP), a disturbance metric to quantify the impact of assignment constraints, and a transparent rationale 'chain of custody' that supports decision-making. The methodology not only enhances assignment match but also provides actionable insights for leadership, enabling more effective talent utilization across the force.

This presentation will detail the modeling approach, data architecture, and optimization logic, offering a scalable blueprint for future Army-wide implementation of decision models to inform AI-enabled decision support tools.

51 sec read

As the DoD and Army continue to characterize controlled weapon systems, understanding turret movement given a target location as an input is essential for calculating Probability of Hit. To achieve this objective, DEVCOM Analysis Center collaborated with Pittsburgh University and together performed accuracy analysis of a controlled weapon system model for differing targets. The government-university team developed a pragmatic mathematical model. This turret model is an input to design two controllers to accept target location input and move a turret to the centroid of the target in simulations. The analysis identifies the factors affecting turret movement, the impact of those errors, and how these errors can lead to incorrect simulations of system accuracy. Further, the team describes the results that support the hypothesis that: turret movement errors are larger over longer distances; accuracy improves the longer a user waits to fire; and that an integral control component is desired to increase accuracy for moving target scenarios. This work directly leads to enabling of comparison of AI to humans.

44 sec read

This study uses fixed and variable video game types to measure pretest sensitization as a proxy for repeated and varied threat test scenarios in system performance testing of air and missile defense systems. The pretest sensitization phenomenon exists when repeated exposure to a test condition influences the participant's response. Research shows air and missile defense development correlates with video games, resulting in similar interfaces and computer operating environments. Department of Defense acquisition test and evaluation results must reflect system performance without prior knowledge of the threat scenarios confounding the results. System performance results inform acquisition decisions, such as further funding and development, program canceling, and fielding decisions. Five video game titles are sampled for each type, using the time(s) to complete the game [first completion(s) versus replay(s)] to detect a decrease in the time to complete the game with repeated exposure, indicating pretest sensitization. The

1 min 52 sec read

As intelligence analysts, we play a critical role in wargaming by emulating the threat and providing a realistic representation of adversary capabilities, doctrine, and intentions. We leverage this role to help Blue gain valuable insights into military operations, develop new understandings of complex scenarios, and inform decision-making. This presentation will explore how our role as "Red" team members contributes to the development of effective wargames, enhances Blue's understanding of itself and the operational environment, and informs Army operations and decision-making.

Intelligence analysts who understand the threat are critical to wargame planning and preparation. By involving Red early in the wargaming process, we can ensure that Blue is presented with a thinking enemy that understands Red's real-world capabilities, warfighting doctrine, and national security imperatives. This allows Blue to test its multi-domain theories, identify potential vulnerabilities, and develop effective countermeasures. Our role as Red team members is not only to present a realistic threat, but also to relay uncomfortable truths to Blue, highlighting potential weaknesses and areas for improvement.

Through our participation in wargames, we can help Blue develop a deeper understanding of the adversary's approaches to warfare, thought processes, and capabilities. This, in turn, enables Blue to develop more effective strategies and tactics to counter the threat. By meshing our knowledge of the threat with an understanding of contemporary and near-future warfare, we can create a more realistic and challenging wargaming environment that prepares Blue for the complexities of modern conflict.

This presentation will draw on lessons learned from various wargaming experiences, including the importance of involving Red early in the game design process, the value of digital tools in augmenting learning, and the need for a centralized location for threat knowledge. It will also be informed by Ian Sullivan's "7 Reflections of a Red Commander, the Army Mad Scientist Initiative's crowdsourcing event and the Game On! Conference, as well as other relevant sources, including Micah Zenko's work on red teaming, Ian Sundstrom's analysis of naval wargaming, and the U.S. Army Pacific's Unified Pacific Wargame Series. By exploring the intersections of wargaming, threat emulation, and military operations, we aim to demonstrate the value of intelligence analysts' contributions to wargaming and highlight the importance of a threat-based approach to Army experimentation and decision-making.

1 min 27 sec read

As the Army faces increasingly contested and complex operational environments, the imperative to become a data-centric force is clear. This work explores how the Army can operationalize data to enable Multi-Domain Operations (MDO) and prevail on the future battlefield, with a focus on supporting senior leaders in translating data-centric concepts into actionable practice.

The approach centers on the principles outlined in the Army Digital Transformation Strategy (ADTS), emphasizing the development of a data-literate culture, standardized data collection, and the integration of advanced analytics—including artificial intelligence and machine learning—across all levels of command. Key to this transformation is the adoption of a Data Enabled Decision-Making (DEDM) framework, which guides leaders through:

• Defining Warfighting Problems with clarity and context, ensuring issues are specific, measurable, achievable, relevant, and time-bound (SMART).

• Identifying and Managing Data Requirements by mapping critical data sources, ensuring data quality, and establishing secure, interoperable data management practices.

• Applying Multi-Level Data Analysis—from descriptive and diagnostic to predictive and prescriptive analytics—to generate actionable insights for operational planning and execution.

• Communicating the Data Story through compelling narratives and visualizations that drive understanding and alignment across the force.

• Upholding Data Governance through VAULTIS Principles (Visibility, Accessibility, Understandability, Linkage, Trustworthiness, Interoperability, Security) to ensure ethical and effective data governance.

• Operationalizing Data by embedding data-driven practices into daily operations, fostering collaboration, and continuously monitoring and evaluating outcomes

By leveraging these methods, senior Army leaders can transform their organizations into adaptive, data-centric teams capable of rapid, informed decision-making across domains. This abstract provides a roadmap for integrating data-centricity into Army operations, equipping leaders to harness the full potential of data in support of mission success on the future battlefield.

2 min 3 sec read

Background: The Wisconsin National Guard, Dept of Military Affairs CHW Program, Basic Course is a two-week course that began in 2017. Early iterations of the course were geared only at physical fitness to recondition flagged Soldiers, however, over time the curriculum integrated more holistic health modalities to meet individuals' unique needs. Objective: The Basic Course was evaluated to determine the effectiveness of the curriculum and how various measures contribute to understanding the impact of course materials to improve overall health and wellness and instill meaningful behavior change.

Method: Using a mixed-methods approach, data from 485 active U.S. Army and Air Force National Guard participants across twelve CHW Basic courses 2017-2024 were analyzed. Sources included aggregated pre-post surveys, individual difference measures and interview transcripts. Surveys addressed individual background and lifestyle factors related to the Army's Global Assessment Tool and five CHW Fitness Pillars: Physical, Mental, Social, Spiritual and Financial. Individual difference measures included the Adverse Childhood Experiences (ACE) scale and the UCLA Loneliness Scale, which were used to identify Soldiers at greater risk for depression, anxiety, post-traumatic stress disorder (PTSD) and harmful behaviors. Pre-post course and yearly mean score differences were compared. Post-course interviews were transcribed and reviewed for key themes.

Results: The survey questions with the greatest statistically significant mean post-course rating score increases were "I know my purpose in life", "I am self-motivated", "Changing my life is hard and nearly impossible*", "I feel stress from my financial situation*", "I have positive self-esteem", and "I often feel down and hopeless*". These items showed consistent stability across years. Individual difference measures revealed 24%-62% of participants [44% overall] had a clinical ACE score ≥4, suggesting adverse childhood experiences and greater susceptibility to adverse adult health conditions. UCLA Loneliness scores showed significant improvements, both pre-post course and across years, indicating decreases in feelings of isolation and loneliness. The CHW fitness pillar, with the greatest statistically significant pre-post course increase was mental fitness, though increases in spiritual, social, physical, and financial pillars were also observed. Participants shared that the course strengthened their mental resiliency and fostered individual mindset changes towards healing and overall health.

Conclusion: These findings indicate that the CHW Basic course positively impacted participants with significant impacts across several health and wellness areas. Many participants reported in interviews that the course changed the trajectory of their lives.

Implications: Though not diagnostic tools, individual measures can help quantify the impact of training, inform development of curriculum content, and ensure support systems during the course.

57 sec read

Abstract: The Army lacks a model to predict equipment combat readiness before, during, and after rotations at the National Training Center (NTC). Reductions in equipment readiness rates can present challenges for Units when completing training requirements. As the Functional Data Manager (FDM) for Predictive Logistics (PL), the Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) is developing a predictive model that will improve Unit planning for training rotations at NTC. The model predicts the potential daily Fully Mission Capable (FMC) % for vehicles during the timeframe starting 60 days prior to the beginning of the NTC rotation, during the rotation, and 30 days after the rotation has completed. Predictions are made at the aggregate Line Item Number (LIN) level. The model applies to statistical techniques using historical daily FMC rates and other data elements from previous NTC rotations to calculate the predictions. A cross-validation approach was used to validate the model. DAC plans to deploy the model within the Army Vantage platform to allow easy access for end users. This model will provide Unit commanders with early insights into the potential variability of FMC rates before, during, and after planned NTC rotations.

36 sec read

Abstract: The FORGE (Force Requirements Generator) model is part of a larger family of applications used for campaign-level force structure capability demand analysis which plays a vital role in the Army's force structure planning and decision-making processes. The Center for Army Analysis uses the FORGE model to determine the mix and quantity of enabler unit types required for each major Defense Planning Scenario in support of Total Army Analysis. By leveraging FORGE for scenario modeling and sensitivity excursions, the Army can make more informed decisions about its composition and structure, ultimately contributing to more efficient and effective military operations. This presentation will focus on the methodology used to determine the supporting Army structure necessary to enable maneuver forces.

1 min 12 sec read

Abstract: Emerging aerial threats and advancements in enemy precision fires have increased the need for effective maneuver air defense (MAD) at tactical echelons. To inform modernization decisions, Office of the Secretary of Defense, Cost Assessment and Program Evaluation directed analysis from the Army and Marine Corps to develop optimal recommendations for MAD capability mixes across various time horizons and threat environments. The Army directed The Research and Analysis Center (TRAC) to perform the analysis for the Program Decision Memorandum task.

TRAC conducted the analysis using a two-phased approach. Phase 1 applied a Mixed Integer Program, built in the General Algebraic Modeling System (GAMS), to identify feasible and cost-effective MAD capability mixes. The GAMS model accounted for a broad array of candidate systems and constraints across the 2030 and 2035 timeframes, while deliberately avoiding over-fitting to any one scenario to maintain relevance across varying conditions within the Indo-Pacific Command (INDOPACOM) Area of Responsibility. Phase 2 involved evaluating the high-payoff MAD mixes from Phase 1 in high-fidelity operational combat simulation environments, which include the Combined Arms Analysis Tool for the 21st Century, One Semi-Automated Forces, and the Rapidly Adaptable Model in Python. This phase focuses on quantifying operational effectiveness in terms of force protection, survivability, and mission accomplishment in notional INDOPACOM scenarios.

This presentation outlines TRAC’s combined optimization and simulation approach to maneuver air defense modernization, highlighting key trade-offs and insights relevant to the Army’s strategic goals in the Indo-Pacific.

43 sec read

Autonomous air vehicles (AAVs) subjected to partial observability and intermittent communications must make sequential decisions under uncertainty while performing a team search and service mission. We formulate a Markov based sequential decision process model of this autonomous aircraft search and service (A2S2) problem and solve it using a blended optimization and reinforcement learning based approach. The team of AAVs must collectively locate and service multiple, moving ground-based targets. We develop a centralized training, decentralized execution (CTDE) multi-agent reinforcement learning (MARL) algorithm that periodically conducts re-planning via a mathematical program, informing subsequent AAV decisions. MARL policies generated by our algorithm are compared to those generated by independent deep Q-networks (IDQN). Designed computational experiments investigating several problem instances show differences with respect to quality of solution, computational efficiency, and robustness over varying mission conditions. Preliminary experimentation suggests that our novel algorithm outperforms established MARL policies.

39 sec read

An ORSA's first assignment is often a whirlwind of new experiences: a higher organizational echelon, unfamiliar unit culture and lingo, technical challenges, and building local relationships in a new environment. This presentation offers a candid look at navigating an ORSA's first assignment, grounded in the author's own assignment at U.S. Army Recruiting Command – their first as an ORSA and at a higher headquarters level. From breaking all the links in a PowerBI dashboard, to designing and teaching classes for a new warrant officer MOS, to shepherding a major TDA-shaping project from concept to approval, the lessons learned apply to ORSAs both new and old. Through personal vignettes, this presentation shares what the author wishes they'd known on day one – and what every new ORSA should hear before theirs.

1 min 14 sec read

A hallmark of operations analysis (OA) in the Modeling and Simulation (M&S) ecosystem is the trade-off between model fidelity and computational runtime. When it comes to employing systems models in the scale required for OA, compromises need to be made between representing systems with full accuracy and keeping simulations times short enough to be run in analytically useful time-frames at scale. Often this involves employing simplified predictive forms of critical system models to reduce runtime while minimizing loss of fidelity. One critical instance of this is lethality modeling, where traditional representations in OA simplify round fragmentation and target vulnerabilities to functions of burst/target distance, or simply a static lookup table probability of defeat. These models, while computationally fast and scalable, lose much fidelity in fragmentation and target asymmetries. Here, we present the development of a novel approach to lethality modeling for operations analysis: The Spatially Partitioned Radial Surrogates (SPRS) approach. Through the application of spatial partitioning, geometric transformation, design of experiments and machine learning techniques, the result is a lethality modeling approach with run times comparable to traditional lookup tables that also maintain radial asymmetry of the fragmentation and target vulnerable areas. As munitions become more precise and exquisite, it becomes increasingly important to improve lethality fidelity in operations modeling to appropriately represent novel systems. At the cost of upfront data generation, SPRS provides a lethality model with much greater fidelity at negligible cost to runtime with the potential to greatly enhance OA capability for the Army.

55 sec read

Mobile nuclear power systems and their applicability to military operations are being investigated across the Department of Defense (DoD). The energy produced by these systems could be used to support a wide array of power needs for the Joint force targeting the single to multi-megawatt range. The Army is interested in developing use cases for mobile nuclear power systems and assessing their performance relative to current power-generation options. DEVCOM Analysis Center (DAC) is leveraging power generation M&S capabilities in conjunction with a library of electrical loads to explore operational use cases and analytically differentiate the performance of mobile nuclear power systems relative to other power generation options. Power generation and the impact on logistics requirements (e.g., fuel demand) constitute the primary analytical trade space, but other metrics, such as thermal-energy production for austere cold climates, maintenance man-hours, and the space footprint required to operate safely, are also being considered. The goal of the effort is to quantitatively demonstrate world-wide locations where mobile nuclear power systems may, or may not, be suitable relative to current power systems available to the Warfighter.

18 sec read

How can models be applied in physical (analog) or computer-assisted wargames? How are models for wargames different than models used in simulations or other analysis? Are commercial wargame models relevant to professional wargames? This presentation will cover practical examples and lessons learned from analytical and educational wargames from the U.S. Army War College's Department of Strategic Wargaming.

51 sec read

Mobile laboratories are deployed to support Combatant Commanders to analyze unknown field samples to counter chemical, biological, radiological, nuclear, explosives (CBRNE) and weapons of mass destruction (WMD). The data generated in these mobile laboratories is laboratory grade multi modal data, that can provide actionable information. Currently, this data and its analysis is not incorporated for upstream analysis to facilitate in the decision-making echelons of multi-modal operations. Advances in machine learning (ML) has enabled data scientists to develop and deploy models that feed into system- of- systems (SoS) providing data aggregation, predictive analytics, and autonomous decision-making. Here we discuss the design of deep learning (DL) models and an automated workflow to predict a chemical structure from a spectrum with results feeding into a chemical knowledge base. The resulting algorithms and workflow allows for robust analysis of unknown samples without depending upon downstream analysis using a database matching. Future work seeks to connect the resulting identification and knowledge based to other data fusion efforts that inform multidomain operations of CBRNE and WMD hazards.

59 sec read

Space continues to evolve as civilian, government, and military interests all deploy satellites and conduct space operations to support a variety of land, sea, air, and space-based missions. Tracking satellites becomes increasingly difficult as satellite numbers grow quickly, with more launch opportunities at cost than ever before. While satellite numbers grow quickly, it takes significant time and resource investment to grow sensor capability to track space-based objects. Maintaining Space Domain Awareness (SDA) is paramount as space evolves. The US Department of Defense historically relied upon the Space Surveillance Network (SSN) to track objects.

The Joint Commercial Operations (JCO) cell within US Space Command (USSPACECOM) expands SDA by purchasing observations from a variety of commercial sensor service providers and aggregating that data in the Unified Data Library (UDL), with nearly 4 billion records input monthly. This non-classified data enhances the SSN, and enables distribution to allies and partners for collaboration in support of operational and strategic objectives. This research aims to analyze and subset that UDL data to create informative metrics and visualizations that support both the JCO and USSPACECOM general SDA as well as arming decision makers with comparative analytics to support future contract negotiations and purchasing decisions.

58 sec read

Quantifying the Impact of Spiritual Readiness Training on Soldier Resilience: A Statistical Analysis of Eagle Mindset in the 101st Airborne Division (AASLT)

The 101st Airborne Division (AASLT) developed Eagle Mindset (EM), a spiritual readiness curriculum IAW FM 7-22, to enhance Soldier resilience and reduce harmful behaviors. This study analyzed EM's effectiveness through statistical metrics collected from over 4,000 Soldiers. Specifically, a quantitative analysis was conducted comparing Commander's Critical Information Requirements (CCIR) events between trained and untrained Soldiers. Out of 65 CCIR incidents recorded within a Brigade of 4,277 Soldiers, EM-trained Soldiers experienced only five events (incident rate of 0.156%), compared to 60 events among untrained Soldiers (incident rate of 5.56%). A chi-squared analysis confirmed statistical significance, demonstrating a substantial reduction in harmful incidents. Additionally, analysis yielded a Number Needed to Train (NNT) of 19, indicating that training 19 Soldiers prevents approximately one CCIR event.

The incident rates for trained and untrained groups were compared using a two-sample independent t-test, yielding a highly significant result. This strongly suggests that Eagle Mindset training significantly reduces incident occurrences, with observed incident rates of 0.156% (trained) versus 5.56% (untrained). The extremely low p-value underscores a robust statistical difference unlikely attributable to random variation alone.

1 min 55 sec read

Abstract: The effective utilization of data is paramount for achieving decision dominance. Historically, data remained siloed in Army Systems of Record and did not easily integrate with internal systems and trackers. The Army has implemented Microsoft’s Power Platform at the lowest echelons to bridge the gap between data and knowledge, empowering decision-makers to leverage both Army and organizational data. The 1st Multi Domain Task Force (1MDTF) Data Team harnesses the resources provided to gather, transform, and apply previously siloed data to enhance operations and decision-making at the tactical level. This presentation will delve into the integration of multiple systems to provide operational value, ensuring that data collected can inform future projects and operations.

How does 1MDTF use data to support a forward presence in the first island chain while maintaining Soldier readiness and accountability? 1MDTF continually rotates Soldiers forward for exercises and sends leaders on TDY assignments to shape future operations and drive innovation. Providing the commander with an accurate operational picture of personnel movements in time and space requires hundreds of work hours each month, and planning future operations demands significant resources to process disaggregated data. Systems like DTS and IPPS-A are used by the DoD and Army to track personnel and travel as official records, while the unit maintains additional internal trackers to process relevant information. Despite the similarity in data types, each of these systems traditionally operates with its own data sets.

Utilizing Army-provided resources, the 1MDTF Data Team is integrating siloed data sources to enhance operational understanding and personnel management across the Task Force. The 1MDTF Data Team works with various staff proponents to develop systems to improve business processes. During development, the 1MDTF ensures that the systems can operate independently while still feeding into each other. The integrated systems include the PERSTAT, the Flight Tracker, and the TDY Tracker. The PERSTAT captures personnel accountability from the lowest level and aggregates it into a single repository. The Flight Tracker is used by CUOPs to track individual movements during travel. Finally, the TDY Tracker consolidates TDY requests to provide visibility of personnel movements and funding. The 1MDTF Data Team achieves this integration through data connections to Vantage, Power Apps, Power Automate, Power BI, Microsoft Lists, Microsoft Forms, SharePoint Libraries, and Army Systems of Record.

1 min 0 sec read

Abstract: Recent global events underscored the strategic capabilities presented by unmanned aerial systems (UAS) in combat scenarios, highlighting a need to understand their vulnerability and improve the protection of future soldiers and ground vehicle systems. To explore the vulnerability of drones to counter-UAS (C-UAS) systems and address this analytical gap, the U.S. Army Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) developed a virtual modeling and simulation (M&S) capability to complement limited experimental data.

DAC developed a high-fidelity finite element model (FEM) derived from a commercial-grade quadcopter UAS. As part of model development, DAC conducted a sensitivity study on mesh density and connection methods of the propeller blades to refine model construction. Using the developed model, DAC simulated experiments related to UAS vulnerability from blast effects, comparing component damage from experiments to measured simulation stresses as an initial verification and validation (V&V) of modeling capability.

This presentation will discuss the model development and validation process, along with its application in supplementing existing experimental data and future empirical models. DAC’s modeling efforts are critical for the Army to assess UAS vulnerability and C-UAS capabilities amid the threat of rapidly evolving technologies.

1 min 40 sec read

Abstract: The Vice Chief of Staff of the Army directed Headquarters, Department of the Army (HQDA) G-8 to conduct the Army Tactical Fires Analysis (ATFA) to assess the requirements for the United States (U.S.) Army’s tactical fires capability and capacity. U.S. Army senior leaders will use this analysis to provide the analytical underpinning to develop the strategy to organize, equip, and modernize its tactical fires platforms and munitions, unmanned aerial system (UAS), and launched effects (LE) capabilities. This effort leverages results from the recently completed Tactical Fires Study Phase 1 (TFS1), Tactical Fires Study Phase 2 (TFS2), the UAS/LE Analysis, and the Center for Army Analysis (CAA) led Cannon Artillery Modernization Options (CAMO). The ATFA results inform investments across the Army transformation continuum and key programmatic decisions for future tactical fires capabilities.

HQDA G-8 tasked The Research and Analysis Center (TRAC) to conduct the analysis for ATFA in two phases. In Phase 1, TRAC considered past analyses efforts from TFS1, TFS2, UAS/LE, CAMO, and other relevant analytical documents to identify analysis that already addressed key aspects from each of seven ATFA lines of effort (LOE). In Phase 2, TRAC focused on answering questions from the seven LOEs that required further analysis, which the team identified in Phase 1.

Due to time constraints to complete Phase 1, TRAC analysts needed to be able to review a large collection of documents in a short time period. TRAC analysts leveraged the Large Language Models (LLM) from the Project Convergence Literature Review (PCLR) tool to conduct a systematic review of the breadth of analytical documents to answer questions across each ATFA LOE. PCLR proved to be an adaptable and effective tool for systematic reviews and may shape the way TRAC conducts systematic reviews and research in future analysis efforts.

This presentation outlines the innovative solution the analysis team used to conduct its systematic review including the steps the team took to prepare and present this information to U.S. Army Senior Leaders.

47 sec read

Recognizing that the Army Materiel Command (AMC) does not have a robust modeling approach to predict and influence unit-based readiness for the fleet of tactical vehicles that will be vital in future conflicts, the AMC Analytics Group (AMCAG) sponsored an innovation competition to solicit solutions to this problem. The objective was for submissions to provide the means for making readiness predictions for armor assets over specified time horizons by battalion. A team from The Research and Analysis Center (TRAC) won second place with a solution that not only made readiness predictions but also made predictions for the repair parts that would be needed to improve future readiness. This presentation will describe the methodology and modeling approach for solving this problem. Key areas include describing the supervised machine learning algorithm and modeling technique, the novel approach for understanding how the Army generates the demand for repair parts, and the potential application of the modeling approach beyond the innovation competition.

55 sec read

Functioning energy infrastructure on military installations is essential in executing daily operations. From powering ranges to maintaining communication systems, energy is a driving factor, and a failure has dramatic consequences. The research conducted by the Energy Resilience team at the United States Military Academy is directed toward enhancing the resilience of the Army's energy infrastructure through a heuristic optimization approach. Deliverables include innovative, sustainable, and actionable design solutions that address resilience, cost, and emissions. Improving resilience, or how quickly a system can return to a satisfactory state after a failure, is the primary focus. The research presented will be the analysis of the heuristic optimization method, recommendations to the algorithm, and current progress. The model incorporates a genetic algorithm to produce a Pareto frontier of micro-grid configurations. The genetic algorithm is coded in MATLAB and works to optimize the defined fitness function. The current results of the model analysis have led to recommendations for implementing adaptive mutation and elitism and adding resilience to fitness functions. The model produced will provide insight into improving energy infrastructure on military installations with a focus on resilience.

2 min 30 sec read

The Army Talent Attribute Framework (ATAF) is a standardized framework consisting of knowledge, skills, abilities, and other characteristics (KSAOs) required by Army personnel and represents a major step forward for standardizing Army talent management processes across the Force. The past research to develop and refine the ATAF thus far has been extensive. However, there is still room for improvement within the framework. As prior research efforts have identified (e.g., Royston et al., 2023), there is meaningful redundancy in the definitions between certain sets of ATAF KSAOs. Additionally, the large number of KSAOs contained in the ATAF can make it difficult for end-users to efficiently and effectively utilize the framework to support talent management processes and programs. This effort builds on past work and offers a more stringent approach for quantitatively and qualitatively reducing the ATAF to a more parsimonious state and maximizing its utility for guiding Army talent management initiatives.

We used a four-phased approach to compile recommendations for ATAF KSAO revision and removal. The first phase consisted of identifying KSAOs that were (a) not measurable KSAOs or (b) clearly redundant with other KSAOs within the framework. The second phase focused on quantitative methods for reduction that included cut-score analyses and factor analyses. The third phase was an in-depth qualitative review which took the recommendations from the first two phases into account when compiling a comprehensive list of recommended reductions and revisions. Finally, ARI Researchers and Army talent management SMEs reviewed the list of recommendations and provided additional feedback on reductions and KSAO content revisions.

In the first phase, 23 KSAOs were recommended for removal after the initial qualitative review. In the second phase, seven KSAOs were recommended for removal based on evidence from cut-scores and 12 KSAOs were recommended for removal based on results from a series of factor analyses. In the third phase, the in-depth qualitative review took the recommendations from the previous phases while also holistically reviewing the framework. This resulted in 81 KSAOs recommended for removal. ARI and Army SMEs then provided relevant feedback that resulted in the adjustment of recommendations where appropriate. Sixty KSAOs were recommended for removal after these reviews, and three new KSAOs that captured key content missing from the ATAF were added to the framework. After implementing the final list of recommendations, the revised ATAF has seven Talent Domains, 33 Talents, and 137 KSAOs. Sixty-eight of the final 137 KSAOs underwent content revision (e.g., updated technical definition/user description, updated title, or moved to a new Talent) based on ARI and Army talent management SME feedback.

The updated ATAF will improve the development of policies, programs, and processes within an integrated and efficient talent management system. The updated ATAF should be incorporated into the Assignment Interactive Module 2.0 (AIM2) Marketplace and used to guide the development of new assessments for use throughout the Army's talent management processes. The revised ATAF was approved by ARI leadership and the United States Army Combined Arms Center (CAC) for use in current and future Army talent management initiatives.

1 min 12 sec read

The Department of Defense Modeling and Simulation Community (M&S) is dependent upon high quality, consistent, and validated Characteristics and Performance (C&P) data to provide actionable insight to leadership. This data needs to be consistently produced, automatically ingestible, and responsive to quick turn analytic needs with minimal programming updates. The U.S. Army Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) has the mission to deliver this data to the M&S community to support Joint Services' analysis. To facilitate this mission, DAC manages the Electronic Data Request System (eDRS), Standard Nomenclature Database (SND) and Army Performance Estimates Database (APEDS). These systems are built in a relational database and feed a web portal interface called the C&P Portal that enables M&S customers to request data products based upon red and blue force structures.

The DAC C&P Data Mission organizes the community to develop requirements, business processes, data standards, and methodologies to deliver critical, usable data to support a variety of analysis across the joint community. This process leverages data science, testing, intelligence, modeling, and analyst expertise to deliver adaptive, but standardized products. The web portal interface has been enhanced to enable automated equipment uploads and data pulls, along with a performance data catalog to bridge knowledge gaps. This briefing will cover the DAC C&P Data Mission processes, enhancements, and web interfaces available for gaining access to the quality, consistent, and validated C&P data.

40 sec read

The Research and Analysis Center (TRAC) have continued to advance its capabilities to model contested logistics and other sustainment functions in the future operating environment. TRAC's Logistics Battle Command (LBC) model is a discrete event simulation written in Java that uses the Naval Postgraduate School's Simkit Application Programming Interface (API). LBC is used by analysts primarily to support Analysis of Alternatives (AoA), i.e., a comparative assessment of weapon systems, vehicles, or concepts of support with respect to their impact on military logistics. LBC also includes the ability to represent maintenance and medical operations.

This presentation will detail LBC's core capabilities and describe the suite of tools included with the model, such as the LBC post processor, database, and a newly enhanced scripting interface that allows analysts to rapidly prototype and evaluate novel logistical concepts.

1 min 0 sec read

To maintain military superiority, the U.S. must adopt a human-hybrid (neo-centaur) approach that leverages Generative AI's (GenAIs) strengths while mitigating the risks of cognitive atrophy and formulaic decision-making. Army leaders, across the levels of war, are aggressively pursuing generative AI solutions to adapt and survive within the evolving battlespace. An awkward, distributed AI-arms-race is developing with technical expertise residing primarily in the private sector. Profit-seeking incentives, rather than a clear-eyed objective for military use, is flooding the commercial sector with spill-over into the tactical force that obfuscates the practical utility of AI technologies for military application. This analysis clarifies the GenAI risk within Army Headquarters, identifies current systems in use, and offers a "centaur" model to talk pragmatic implementation of AI systems within military organizations. Ultimately, Army leaders must: (1) avoid the natural proclivity for automation bias ("minotaur drift") and (2) create hybrid-organizations, implementing AI systems with intentionality to enhance the Army's warfighting advantage.

This concept is currently being tested in wargaming. Most recently, the SCALE AI system Donovan was used to support decision making during the ARNORTH warplans wargame. Future wargaming will include support to the National Security Council crisis exercise run by the Basic Strategic Arts Program (BSAP).

55 sec read

Imagine being able to model complex systems, predict outcomes, and inform critical decisions with unprecedented speed, scalability, and collaboration. Welcome to the world of cloud-based discrete event simulations (DES), where open-source Python solutions are transforming the way organizations approach simulation modeling. In this presentation, we'll explore the game-changing benefits of running DES in the cloud, breaking free from the limitations of traditional closed networks and single computers. With cloud-based DES, you'll gain access to premier tools for collaboration, scalability, security, and usability, enabling your team to work more effectively and make data-driven decisions with confidence. But don't just take our word for it - we'll share a real-world use case that demonstrates the power of cloud-based DES in action. Learn how we used DES to evaluate the Army's ability to perform Large Scale Mobilization Operations, providing detailed insights and enabling Course of Action development and analysis that inform operational decision-makers. Join us to discover how cloud-based DES can transform your approach to simulation modeling and decision-making. Get ready to drive meaningful change with the powerful insights and results that cloud-based DES can deliver.

57 sec read

The Army has used military value analysis (MVA), a decision-analysis method, to aid in stationing decisions starting with Base Realignment and Closure (BRAC) 2005 and continuing through 2020 for stationing of brigade-sized units. Over time, at the request of Army senior leaders, MVA incorporated treatment of non-contiguous training areas and added community support value analysis (CSVA) to consider an installation's off-post quality of life and community support. Department of the Army Pamphlet (DA PAM) 5–13 (Procedures for Army Stationing, 2 April 2025) outlines the Center for Army Analysis' (CAA) role in this process: "[CAA] conducts the quantitative analysis of the MVA/CSVA model as required by the DCS, G–3/5/7 FM." From 2009–2020, CAA supported annual G–3/5/7 (DAMO–FM) requests for decision support for stationing actions. On 21 March 2025, the Director, DAMO–FM, requested CAA conduct MVA and CSVA to inform Phase III (Synchronization) of Total Army Analysis (TAA) 28–32. DAMO–FM provided a memo outlining the installations under consideration, desired attributes with associated weights, and requested addition of four factors for integration into stationing analysis. This briefing will cover the background, installation attributes considered, and the process used to conduct MVA and CSVA for Army stationing decisions.

53 sec read

Abstract: The U.S. Army Operational Test Command (USAOTC) has initiated support for Expeditionary Operational Assessments (EOAs) to assist the Army Evaluation Center (AEC) in conducting early evaluations of emerging and commercial off-the-shelf (COTS) technologies in operational environments. These assessments aim to accelerate the delivery of innovative capabilities to Soldiers by integrating user feedback early and often. EOAs are typically conducted with Transformation in Contact (TiC) units during Combat Training Center (CTC) rotations, providing valuable Warfighter insights on equipment effectiveness in realistic scenarios.
In October 2024, USAOTC was tasked with executing an EOA of the Next Generation Tactical Vehicle (NGTV) during the January 2025 CTC rotation at the Joint Multinational Readiness Center (JMRC) in Hohenfels, Germany.

This presentation outlines the key challenges faced during planning and execution, including limited planning factors for NGTV employment, the design of the data collection strategy, and mid-assessment adjustments required due to the dynamic nature of EOAs. It will also highlight the methods used to analyze Warfighter feedback and system utilization, as well as discuss notable shortfalls encountered by the test team.

58 sec read

Abstract: Raman spectroscopy has recently emerged as a promising technique for bacterial sample identification, with potential applications in cellular analysis, disease diagnosis, and microorganism differentiation. However, Raman analysis of complex samples is currently challenging because bacterial spores and other microorganisms are difficult to distinguish from the background surface under a microscope for complex surfaces, hindering accurate Raman laser focusing. To address this, we present an illumination strategy that enhances biological sampling on complex surfaces by improving sample-to-surface contrast, using aerosolized bacterial spores as a model. This strategy employs a 475 nm LED, coupled with a 520 nm bandpass filter positioned in front of the microscope camera, to stimulate intrinsic flavin autofluorescence in bacteria. Consequently, bacterial spores larger than three micrometers were readily visible on diverse complex surfaces, even those exposed to ambient environmental conditions. However, the utilized illumination strategy suffers from limited illumination depth and the inability to identify particles smaller than three micrometers, limitations we aim to address through future funding. These findings establish a crucial framework for integrating artificial intelligence and machine learning with a portable microscope developed at the Chemical Biological Center, ultimately enabling automated identification and analysis of potential microbial threats.

56 sec read

Abstract: Modeling and simulation (M&S) is often utilized to explore concepts, assess capabilities, and inform decisions. As the Army modernizes and transforms, the need to quickly identify, modify, configure, and integrate M&S to address specific questions become increasingly important. The Army Modeling and Simulation Office (AMSO) established an unclassified M&S Enterprise Innovation Environment (MSEIE) with the US Army Engineer Research and Development Center (ERDC) via the ERDC Cloud Computing Environment to facilitate configuration, modernization, integration, and testing of migration to Modular Open Systems Approach (MOSA) based M&S. The MSEIE provides authorized users across the Army M&S enterprise and industry partners access to M&S that has been targeted for migration to MOSA as well as to development, security, and operations (DevSecOps) tools. Simulating Robotic Breach, a pilot project tested within the MSEIE, was implemented using higher-fidelity ground vehicle robotics simulation, OneSAF, and integration protocols. This presentation will provide an overview of the MSEIE, including architecture; describe the models, simulations, integration, and results; relay lessons learned from the Simulating Robotic Breach pilot project; and identifying next steps and opportunities for future work.

1 min 14 sec read

Contested logistics consists of the efficient and safe transport of personnel, materiel, and services across an array of modes or lines of communication (LOCs), while overcoming adversarial-induced multidomain effects intended to impede Army and Joint Sustainment operations. Movement control principles facilitate the allocation, coordination, and routing of transportation assets in accordance with command priorities, encouraging spatiotemporal synchronization of logistical assets and units across intra-theater LOCs. This spatiotemporal awareness provides logistical solutions that answer how (i.e., planning and routing) and in what manner (i.e., scheduling route movements) units need to move through contested operational environments to overcome enemy effects and support tactical front movement. However, complex and dynamic logistics networks along with other operational factors increase logistic route planning uncertainties, impacting the warfighter's ability to make well-informed, timely decisions. This research focuses on developing a decision support capability for planning, development, and selection of multimodal routes within transportation networks in support of distributed logistics applications in order to 1) provide viable sets of multimodal shortest path (SP) solutions within contested logistic environments, 2) incorporate user preferential route data, and 3) deliver route and distribution metrics for course of action development. These efforts support providing the Army with an automated routing capability that empowers routing-focused decision making, while informing other movement control functions in accordance with Army doctrine. This presentation will provide an overview of findings from analyzing similarity between SP solutions by adjusting user preference modifiers, impacts of road geometry on routing, and disclose a novel movement control framework.

58 sec read

As the vast landscape of Large Language Model (LLM) research continues to evolve, one key consideration has remained ever present. In specialized contexts like Army analytics, where resources are constrained and domain knowledge is paramount, commercial LLMs predominantly trained on internet-sourced text can be less valuable than smaller, locally hosted, expert LLMs originating from open-source efforts. In this work, we introduce TRACLM-V4, the next generation of LLMs in the TRACLM project, representing the latest evolution of LLM adaptation to the Army domain. Anchored by TRACLM-DATA-V4, the largest and most comprehensive for Army LLM instruction tuning to date, TRACLM-V4 leverages an unprecedented amount of synthetic data from state-of-the-art open-source LLMs to deliver models of unparalleled domain-specific knowledge. For the first time, TRACLM-V4 models incorporate function calling data, enabling agentic behaviors that can augment analytical workflows in the Army. With models available in various sizes to support everything from edge devices to supercomputers, TRACLM-V4 aims to address a broad range of use cases and deployment environments. This presentation summarizes our processes for training and evaluating TRACLM-V4 models, and we invite the military analytical community to openly experiment with developing innovative applications using TRACLM-V4 models.

52 sec read

This presentation exposes the lessons learned from development of a custom Power Platform tool for US Army Pacific (USARPAC) to assign and track talent in billets across the headquarters. At USARPAC, personnel are often dynamically tasked out to support directorates other than the office to which they are officially assigned. This causes a myriad of issues as leadership struggles to assess true capacity and capability within directorates. The USARPAC Chief Data Officer (CDO) staff was commissioned to develop a set of tools that would assist with tracking individuals as they move around the staff (outside of their force management paragraph and line number). The Personnel and Spaces Alignment Tool (PASAT) consists of interactive Power Apps and Power BI interfaces that are connected with an intricate series of Power Automate and querying functions. This presentation gives an overview of the PASAT success story for taming talent mobility at USARPAC. The focus is to share how it is currently employed while highlighting the challenges of developing, maintaining, and integrating such a platform into daily operations.

1 min 3 sec read

To defend Japan's territory, which span over 3,00km, the MOD/SDF is fundamentally strengthening its stand-off defense capabilities. This will enable diverse responses from outside the threat zone of anti-air missiles and similar systems against vessels and landing forces attempting to invade, including those targeting remote islands. In this regard, the Ministry of Defense has been working on research and development of long-range anti-ship missiles, including surface-launched, ship-launched, and air-launched, and has acquired high technological capabilities, while the technology of ships of other countries is also gradually improving, and they have high interception capabilities. To cope with these issues effectively and efficiently, it is necessary for us to develop a simulation tool that can evaluate attacks on ships by anti-ship missiles.

Decision-making regarding anti-ship combat is becoming more complex due to increase in factors to be considered, such as evaluation of the fleet's interception capabilities in addition to the interception evaluation of individual ships, reduction in the interception capabilities of ships due to electromagnetic interference, vulnerability of ships to anti-ship missiles, evaluation of targeting conditions, etc. As a result, we need simulations that can be performed in a relatively small number of people and in a short time to support commanders in making decisions regarding firing direction, direction, timing, etc.

44 sec read

DEVCOM Analysis Center (DAC) is responsible for generating authoritative ground combat system performance data, to include large caliber direct fire weapon accuracy. As the DoD and Army move toward integrating unmanned aerial systems (UAS), understanding the impact is essential to properly characterize Probability of Kill given a Shot. DAC will showcase the current ability of image stitching, a UAS with the ability to sense cross wind speed and direction, and key values of the error budget associated with reconnaissance.

This presentation will cover a unique sandbox imaging experimental set up to investigate target location error critical to Probability of Hit. This work will highlight the physical UAS that was developed, tested, and flown. The presentation also describes the impact of cross winds for terminal ballistics. The work describes individual errors, and the next steps to include systems engineering analysis with an emphasis on time lag implications.

1 min 38 sec read

Shadows and Steel 2045 represents an example of a new analytic tool that complements rather than replaces existing operations research methodologies. Selected as a winning entry in the Deputy Secretary of Defense/Vice Chairman JCS Academic Research Challenge (AY25), this approach was designed to address a critical limitation in DOD analytic capacity: the narrow participant base for concept development that constrains hypothesis generation and data collection.

While high-fidelity simulations and formal wargames remain essential for detailed analysis, they suffer from participation bottlenecks that limit the diversity of strategic insights available to analysts. This deck-building game enables rapid hypothesis testing across multiple iterations while generating quantifiable data on capability combinations, force structure preferences, and strategic priorities. Through five design iterations and controlled testing with USAWC faculty, DOD students, and service academy participants, we validated that the game's accessible mechanics capture essential strategic tradeoffs while generating foundational analytic value.

Key advantages for the operations research community include: drastically reduced time-to-insight (15-30 minutes vs. multi-day exercises), elimination of facilitator bottlenecks, large-scale distributed data collection, and identification of promising concepts for detailed simulation analysis. Testing demonstrates that participants consistently seek to balance exactly the kinds of variables that lead to optimal force packages and quickly learn to demonstrate sophisticated understanding of capability interactions.

The approach creates a wider analytic funnel: broader participation screens strategic concepts and generates hypotheses that could then be more rigorously analyzed through traditional simulation methods. Rather than replacing existing tools, this expands the toolkit with a scalable front-end that feeds higher-fidelity analysis with better-informed starting points. Moreover, players who achieve mastery of this game may well be demonstrating some of the skills necessary to move into professional concept development and force design positions, increasing the pool of available candidates.

Senior DOD leadership's selection of this approach signals institutional recognition that expanding analytic participation may be as critical as improving analytic precision—offering the operations research community a new tool for scaling their impact.

1 min 14 sec read

Modern tactical battlefield networks are resource-constrained and saturated with devices. Their static designs create an environment that is slow or unable to react to cyber attacks or dynamic changes in resource demands that originate from modifications in the commander's objectives or evolving mission requirements. The DEVCOM Analysis Center (DAC) has the responsibility to perform cyber vulnerability analysis (VA) on these tactical battlefield networks, and DAC requires tools and methodologies that can analyze and assess how resilient these networks are to cyber attacks and vulnerabilities introduced by resource constraints that impact network communication effectiveness.

As part of these assessments, DAC has identified that an opportunity exists to leverage software defined networking (SDN) to address many of the deficiencies above through the use of dynamic network resource control and network slicing through intent programming. The methodology that DAC has proposed to mitigate cyber vulnerabilities could detect a cyber attack or rapid change in resource demand then automatically mitigate the attack and reallocate resources as needed to enable mission success. DAC's assessment of this potential solution during experimentation showed successful mitigation of several cyber attack types, as well as a 35% increase in message completion rates for network devices when employed in Squad and Platoon-sized network environments.

In addition to sharing these results to inform requirements development and system design, DAC will also use lessons learned and implement some of the techniques from this framework into cyber analytic tools to assess and analyze Army networks and systems for vulnerabilities.

1 min 41 sec read

Prevalent assumptions within the US Army often characterize new Soldiers by specific attitudes regarding suicide, substance abuse, finances, and other risk factors or harmful behaviors. Leaders who operate based on these preconceptions may lack awareness of new Soldiers' actual tendencies toward risk and harmful behavior, overlooking valuable opportunities to positively shape their organizational culture with each incoming service member.

This research investigated the influence of initial military training on these risk variables and sought to generally describe the characteristics of the latest generation of Soldiers reporting to their first unit.

In collaboration with the Army Substance Abuse Program, researchers administered Unit Risk Inventory (URI) surveys over 18 months to 495 prospective 35F Intelligence Analysts arriving from Basic Combat Training (BCT). The goal was to identify whether their risk variables were elevated or reduced compared to Army averages. Subsequently, 313 of these individuals completed the same URI after finishing their Advanced Individual Training (AIT) to measure the impact of their initial entry training (IET) on those variables. The study population comprised active component, National Guard, and Army Reserve members ranging in rank from Private to Specialist and spanning various ages.

The study revealed that Soldiers completing AIT were almost half as likely to have experienced suicidal ideation in the preceding 12 months compared to the active Army average. While still lower than the Army average, these Soldiers also showed a reduced likelihood of trusting their chain of command with personal issues. Notably, new Soldiers reported greater life satisfaction and were nearly half as likely to face financial debt or related difficulties. Additionally, Soldiers leaving AIT were approximately 20% less likely to engage in drug use if the Army discontinued random drug testing.

These findings offer a more optimistic perspective on new Soldiers than is commonly held in supervisory discussions. The results underscore the critical role of local commanders and NCOs in gaining units to establish environments that preserve the positive attributes of new Soldiers before their risk variable profiles converge with Army averages related to harmful behaviors.

1 min 40 sec read

Abstract: In the complex and rapidly evolving environment of large-scale combat operations (LSCO), commanders require sophisticated tools to assess operational effectiveness and enable data-informed decision-making. Technological innovations in recent years have enabled greater situational awareness to private industry leaders via rapidly expanding computational power and artificial intelligence to analyze a wide range of data applications. To maintain decision dominance, or at least parity, in LSCO, the Army is leveraging similar capabilities to build common operating pictures (COPs) at the strategic down to the tactical level.

While some units experiment with these costly innovations, most divisions (the principal tactical echelon of the U.S. Army, according to FM 3-0) still adhere to the same staff-intensive assessments process (documented in ATP 5-0.3 last published in 2020) using static reports and PowerPoint-driven running estimates that they’ve used for decades.

This presentation introduces an integrated assessment tool, a modernized evolution of the 2017 Correlation of Forces and Means (COFMs) calculator, using existing Power BI and SharePoint capabilities to enhance situational awareness and tactical planning by integrating daily inputs of friendly and enemy combat power, in addition to projected engagements and targeting of enemy systems. Drawing on insights from two warfighter exercises (WFX) and the 3rd Infantry Division’s journey to becoming a data-centric division, the tool operationalizes data from across the six warfighting functions (WfFs) to quickly update the commander's operational assessment, on demand.

Building on the COFMs calculator force equivalents and loss ratios, the assessment tool processes near-real-time data on friendly and enemy combat power to generate intuitive visualizations of unit readiness projections and resource allocation. By aligning assessments with the division’s planning horizons, the calculator ensures evaluations focus on time, space, and purpose, fostering a disciplined approach to measuring progress at the tactical level in LSCO. Its flexible design supports an adaptive battle rhythm, enabling commanders to synchronize assessments with targeting cycles and respond to volatile conditions and inform division-level planning without the need for expensive built-for-purpose software solutions.

1 min 3 sec read

Abstract: This abstract is part of a broader series of abstracts from the U.S. Army Test and Evaluation Command (ATEC), where we will discuss digital transformation efforts underway in ATEC. ATEC’s senior leaders have challenged our workforce to improve the efficiency and efficacy of the test and evaluation portion of the acquisitions process while reducing live testing by 30% by 2030. Various efforts currently underway provide the components necessary to achieve this vision.

The ATEC Modeling and Simulation (M&S) Hub provides increased opportunities for use of M&S on Army evaluation programs through improved M&S literacy and facilitates new relationships between testers, evaluators and modelers. Also, the hub provides ATEC with greater visibility of their M&S assets that are available for use and identifies gaps in M&S capability across the T&E landscape.

The overview of the M&S Hub will discuss purpose, benefits, current status, and future development plans or the tool. It will also discuss how the repository fits into the greater evaluation framework. The demonstration of the M&S Hub will show evaluators how the M&S Hub repository can be leveraged to identify M&S tools for supplementing or replacing physical testing, and to track tool usage across programs.

52 sec read

Abstract: Operations and Maintenance (O&M) for facilities and installations is essential for maintaining operational readiness and mission capability. However, the available funding to accomplish O&M at facilities and installations usually fails to meet the demand. To optimize the use of available funding, it is imperative to allocate O&M resources through a defensible, objective, and cost-effective methodology. Researchers have developed a Line of Sight (LoS) value model framework to aid in budgetary decision-making across an annual O&M budget of $5 billion for military facilities. The research team has implemented this framework in a case study for multiple domains in support of the United States Army Corps of Engineers (USACE). While the initial case study examines civil-works applications, the framework is versatile and could inform stakeholders for O&M prioritization in military budget planning. This presentation will outline the general LoS value model framework, describe the results of the initial case study for the USACE, discuss lessons learned during the development process, and provide recommendations for adapting the framework in military budgetary decisions.

1 min 7 sec read

This research addresses the critical challenge of operating Army logistics and transportation networks in contested environments by developing a novel approach to modeling and solving the max flow network interdiction problem. Unlike traditional optimization approaches, we formulate this problem as a Markov Decision Process (MOP) to determine the attacker's optimal interdiction plan on a network where interdiction outcomes are stochastic. We then utilize the structure of our MOP to propose a novel extension for modeling attackers with realistic threat behaviors. This approach models suboptimal attackers guided by tactical objectives, framed as tactical tasks (e.g., canalize, turn, or isolate) and incorporated into the reward function of our MOP. In computational experiments, we demonstrate how reinforcement learning (RL) agents trained on our MOP can generate optimal solutions comparable to standard mixed-integer programming solvers for networks with stochastic interdiction outcomes, as well as interdiction decisions aligned with a potential threat's task. We also quantify how attacker intelligence on defender routing impacts interdiction efficacy. Finally, we present a case study demonstrating our adaptive method of training the RL agent to emulate an anticipated threat's objectives and decision patterns. This approach provides a more nuanced understanding of potential network vulnerabilities by considering a range of enemy courses of action from most dangerous to most likely, enabling planners to prioritize hardening routes vulnerable to both optimal and task-driven attacks.

36 sec read

As Large Language Models (LLMs) are increasingly employed to ingest, understand, summarize, and interact with documents, the importance of enhancing their performance through data pre-processing becomes evident. Effective metadata is a key building block for enabling LLMs to understand and query a knowledge domain or an enterprise's document library. This experiment explores methods, tools, and techniques (MTT), including but not limited to leveraging LLMs, for transforming messy metadata into structured and meaningful formats that support accurate and efficient user queries. The performance of various metadata enhancing MTT is benchmarked against human-defined, well-structured metadata, demonstrating a scalable and effective approach to creating reliable metadata. This process ultimately improves LLM performance and reliability when applied to large and often messy document libraries.

25 sec read

The U.S. Army employs a structured yet subjective methodology to evaluate officers for career progression, primarily through a formalized board process. The process balances empirical data with subjective judgment using individual board member assessments, performance-based evaluations, and structured recalibration mechanisms to reward performance and identify officers with superior future potential for promotion to higher leadership positions. This paper examines the mechanics of the board systems, analyses whether scoring is consistent year after year, and if changes should be made as a result.

1 min 44 sec read

Lethality analysis is the modeling and simulation of a weapon and munition pairing against a target. Historically, lethality analyses are conducted in a 2D or 3D grid-like pattern over a space of interest to create a vulnerability map. Other factors of interest, such as terminal velocity, angle of fall, target, kill criteria, and warhead (z-data file) are typically set constant for a given map, and multiple vulnerability maps may be generated if there is interest is seeing the effects of changing those other factors. Results are typically presented as a color-coded visualization of the probability of kill (or other metric of interest) for each point, which might be the average of up to 100 Monte Carlo simulations of that same location/scenario. The set of relevant vulnerability maps may then be used as a lookup table for weapon effects during a follow-on analysis incorporating weapon delivery error. This approach can quickly result in a large computational burden with millions, billions, or sometimes even trillions of runs required as the number of factors of interest increases the spacing needed between points decreases. Here, a methodology is presented that incorporates design of experiments and machine learning with artificial neural networks (ANN) to reduce this computational burden and make accurate lethality predictions over the entire effects space of interest.

Once the ANN model has been fit to the data, the result is a closed form, fast-calculating equation that can be used to predict the mean lethality metric of interest at any point in the design space, not just the specific points that were simulated. If there is a need to create 2D or 3D vulnerability maps, then that can be done with the model predictions. With an appropriately fit model, these predictions should be as accurate or possibly even more accurate than the lethality plot otherwise created from the average of a potentially un-converged number of Monte Carlo runs. And by applying design of experiments techniques, it is likely that orders of magnitude less simulations were required to fit the model than those required to directly simulate the dense grids.

27 sec read

Physics-of-Failure (PoF) analysis provides a mechanism to increase system reliability and achieve test efficiencies. DEVCOM Analysis Center (DAC) engineers continue to support DEVCOM centers, the Army Test and Evaluation Command, and the Engineering Research and Development Center with PoF modeling and simulation to identify design and testing improvements for Army and DoD systems. These system structural and thermal analyses identify design improvements and provide information to augment and streamline testing. This presentation shows examples of how DAC is using PoF to support Army and DoD programs along with the associated benefits.

45 sec read

The Army needs to adapt to a future battlefield that is increasingly transparent, and dominated by highly lethal enemy multi-domain strike complexes. The Future Studies Program (FSP) is the Army's series of Title 10 Wargames that examines the changing character of war and informs decisions on concept development to ensure the Army of 2040 can survive and win against its adversaries. To quantify the ability of Army formations to endure during protracted conflict, The Research and Analysis Center (TRAC) supported the FY 25 FSP events with sustainment modeling using the Logistics Battle Command (LBC) simulation.

This presentation will describe the unique challenges and methods that arose from using LBC to model the future formations, concepts, and environments examined in FSP wargaming. Some key areas of analysis included: the impact of force protection on survivability of sustainment units, the value of pre-staged materiel, and improved distribution efficiency achieved through reduced consumption.

2 min 5 sec read

The identification and characterization of defects within energetic materials are crucial for ensuring the safety, reliability, and performance of warheads. Traditional methods of defect detection often rely heavily on manual inspection of X-ray images, a process that is time-consuming, subjective, and susceptible to human error. This paper explores the potential of machine learning and image processing techniques to revolutionize this critical aspect of warhead assessment. Specifically, this research investigates the application of advanced machine learning algorithms to X-ray images of energetic materials. X-ray imaging provides valuable insights into the internal structure of these materials, enabling the detection of subtle defects that may not be visible through other inspection methods. Our approach leverages both supervised and unsupervised learning techniques to develop robust and accurate defect detection models. Within the realm of supervised learning, we explore the use of binary classification algorithms to distinguish between defective and non-defective regions within X-ray images. Furthermore, we investigate the application of object detection and image segmentation models to not only identify the presence of defects but also to precisely locate and delineate their boundaries within the images. We have also used object detection models to automatically identify image quality indicators in the x-ray calibration process allowing for quicker signal to noise and contrast to noise ratios (SNR and CNR) calculations. To enhance the performance of these models, we experiment with various feature extraction techniques to extract the most informative characteristics from the X-ray data. In the domain of unsupervised learning, we employ clustering algorithms to group similar image regions based on their inherent characteristics. This approach can be particularly useful for identifying subtle defect patterns that may not be easily captured by predefined rules or labeled data. Additionally, we explore the use of data verification techniques to ensure the quality and consistency of the training data, which is essential for building robust and reliable machine learning models. To demonstrate the efficiency of the proposed methods, this paper presents results for a range of performance metrics, including model accuracy, precision, and recall. These metrics provide a quantitative assessment of the models' ability to correctly identify and classify defects within X-ray images. This research directly contributes to the Energetic Defect Characterization (EDC) Capability Development Effort at DEVCOM Armaments-Center, which aims to improve the characterization of defects in energetic materials, and further develop a scientifically verified, evidenced-based predictive capability at Picatinny Arsenal. By automating and enhancing defect detection, this research has the potential to significantly improve the safety, reliability, and performance of warheads.

1 min 22 sec read

Program Evaluation of the Comprehensive Health and Wellness (CHW) Leaders Course and the use of a Mixed Method Approach including Qualitative Data to Inform Program Impact among U.S. Army and Air Force National Guard Personnel.

Background: The Wisconsin National Guard, Dept of Military Affairs CHW program, Leaders Course is a 5-day course that began in 2019. The course is geared toward junior enlisted and above to enhance leadership skills through communication strategies that empower the individual; fosters a supportive environment and culture of growth. The course curriculum centers around Motivational Interviewing (MI) to facilitate behavior change.

Objective: The objective of the analysis was to determine how various measures and types of data contribute to understanding the impact of course components to build leadership skills and enhance the ability to implement meaningful behavior change.

Method: A mixed method approach was used to assess the course. Data from 137 active U.S. Army and Air Force National Guard participants across five CHW Leadership courses conducted between 2019-2024 were analyzed. Sources included aggregated pre-post quiz and survey data, individual difference measures, and qualitative data including focus groups (FG) and homework assignments. The quiz included 27 questions related to the curriculum to gauge participants' foundational understanding of the material. Surveys addressed background, lifestyle and individual difference measures. Pre-post course and yearly mean score differences were compared. FGs were conducted with participants and Cadre to understand their unique perspectives about the course while daily reflection papers served as a place for participants to describe their learning experience and change papers provided an opportunity to outline an issue that they want to change.

Results: Participants scored significantly higher (p<.05) on the post course than the pre-course quiz and this was consistent over the years. The questions with the greatest pre-post change were about MI techniques / principles while the smallest pre-post change were about dialogic change models. Qualitative data from FGs, reflection and change papers include common themes describing the importance of MI skills and implicit and explicit components of the course highlighting that together build Leader self-efficacy, ability to communicate, motivation to create change, a deeper understanding of themselves and their Soldiers and a toolbox of social-emotional skills that are currently not provided in other Military training courses.

Conclusion: The findings describe methods that instill trainable traits, skills and behaviors needed to be an effective, compassionate leader and role model to facilitate culture change. Qualitative data describe how soft skills" aren't learned in the traditional sense

40 sec read

Abstract: Asia Pacific Analytic Platform (AP)² is the Indo-Pacific data sourcing, sharing, and analytics platform that stems from a requirement for sharing unclassified, multilingual PAI and CAI. (AP)² was developed by USARPAC's Chief Data Office to provide a wide range of end-users with government-owned data as well as visualization and analytic tools. USARPAC curates and stores immense quantities of publicly available data; (AP)² enables interaction with this data in a way that is fully accessible (on IL-2 and IL-4 networks) to authorized US and partner-nation personnel in the Indo-Pacific region. The platform is currently being used for operational planning, advanced geospatial analysis, and custom search and visualization on open-source data. This presentation focuses on the lessons learned from one year of requirements gathering, platform development, data sourcing, and leadership of an engineering team.

1 min 35 sec read

The Vice Chief of Staff of the Army directed Headquarters, Department of the Army (HQDA) G-8 to conduct the Army Tactical Fires Analysis (ATFA) to assess the requirements for the United States (U.S.) Army’s tactical fires capability and capacity. The U.S. Army may use this analysis to provide the analytical underpinning to inform the strategy to organize, equip, and modernize its tactical fires platforms and munitions, unmanned aerial systems (UAS), and launched effects (LE) capabilities. This effort leverages results from the recently completed Tactical Fires Study Phase 1 (TFS1), Tactical Fires Study Phase 2 (TFS2), the UAS/LE analysis, and the Center for Army Analysis led Cannon Artillery Modernization Options (CAMO). The ATFA informs investments across the Army transformation continuum and key programmatic decisions for future tactical fires capabilities.

HQDA G-8 tasked The Research and Analysis Center (TRAC) to conduct the analysis for ATFA in two phases. In Phase 1, TRAC considered past analyses efforts from TFS1, TFS2, UAS/LE, CAMO, and other relevant analytical documents to identify analysis that already addressed key aspects from each of the seven ATFA lines of effort (LOE). In ATFA Phase 2, TRAC focused on answering questions from the seven LOEs that required additional analysis, which the team identified in Phase 1. The analysis team also conducted new analysis to supplement past work. The team conducted two tabletop exercises with appropriate warfighter participation to refine vignettes within strategically aligned operational scenarios. The team used several tools, to include combat and optimization models, to conduct analysis across the different vignettes, organizations, capabilities, and mission sets. The team also conducted cost analysis, schedule risk analysis, performance analysis, sustainment analysis, and doctrine, organization, training, materiel, leadership, and education, personnel, facilities, and policy analysis.

This presentation will cover the approach the analysis team developed; the models, methods, and tools utilized; and lessons learned during the preparation and execution of the analysis.

58 sec read

Abstract: The modernization of military installations is imperative to address the increasing complexity and potential risks posed by current and future threats. A critical component of this modernization effort is the transformation of real property master planning, which involves the strategic development of an installation's land, facilities, and infrastructure while ensuring efficient resource utilization and effective risk management. As military installations grow in complexity and hazard exposure, the real property master planning process necessitates the integration of extensive and diverse data sets. These data sets originate from various disconnected and disparate sources, resulting in knowledge silos and substantial manpower requirements to access and compile the necessary information. To enhance the efficiency and effectiveness of real property master planning, researchers are undertaking a comprehensive mapping of the available data and data sources utilized in this process. This data mapping initiative aims to support the modernization efforts by providing a cohesive and accessible framework for real property master planning. This presentation will address the challenges associated with data management in real property master planning, present the current progress of the data mapping effort, and share valuable insights and lessons learned from navigating multiple data systems.

47 sec read

The Army Capability Manager Field Artillery Brigade (ACM FAB) Director requested support to better understand the impacts on the Class V distribution capacity when altering rocket length, personnel reload expertise, and Forward Support Company (FSC) distancing. In a zero growth Army, batteries and FSCs have a finite number of ammunition transport vehicles, trailers, and support assets to satisfy the high demands of ammunition requirements during large scale combat operations. Early engagement to quantitatively analyze design impacts on the existing logistical support system better informs the requirements process prior to a DOTmLPF Assessment of follow on analysis. The Capability Development and Integration Directorate Analysis Branch utilize an operational combat model to underpin mission requirements before overlaying support system variability. This presentation describes the approach of quantifying demands and incorporating a high-level scheduling model to assess varying supply chain architectures. The tradeoff results support the artillery community's interest in assessing the viability, value, and trade-offs while designing DIVARTYs with additional capabilities.

57 sec read

The SOUTHCOM J86 is exploring the potential of Large Language Models (LLMs) to reduce the administrative burden of parsing and extracting standard data from non-standard exercise after-action reviews, a crucial data source for supporting assessments. Currently, the non-standard nature of these reviews, combined with varying content quality, hinders the accuracy and relevance of assessments. To address this challenge, we utilized DoD-developed LLM interfaces, such as NIPRGPT and CamoGPT, to automatically extract information from disparate text-based datasets. Our testing involved prompting the model to extract standard data from both fictional and real-world after-action reviews, with a goal of achieving 70% accuracy. The results showed that NIPRGPT achieved up to 93% accuracy on fictional reports and 83% accuracy on real-world reports, significantly outperforming human analysts in terms of speed and efficiency. Notably, the model processed 10 fictional reports in under one minute and 10 real-world reports in approximately 4 hours, compared to the approximately 3 hours it would take a human analyst to read similar reports. The extracted answers were clear, concise, and faithful to the source material, demonstrating the potential of LLMs to support assessment requirements and reduce the administrative burden associated with non-standard data collection and reporting.

1 min 20 sec read

This study examines the piloting of Behaviorally Anchored Rating Scales (BARS) during the FY26 Reserve Officer Training Corps Professor of Military Science Board (FY26 ROTC PMS Board), representing the first significant departure (since the 1950s) from the Army's traditional word picture used during selection boards. BARS establishes more objective selection criteria designed to mitigate: (1) untested heuristics, (2) illusions of validity, and (3) personal idiosyncrasies that can influence traditional selection outcomes.

Our methodology introduced two innovative approaches: utilizing a Large Language Model (LLM) with Retrieval Augmented Generation (RAG) to identify key performance dimensions in minutes rather than hours as well as employing a novel survey approach with subject-matter-experts to rank performance indicators that generate anchor links within the BARS framework.

Our analysis of data from this 5-day board, which evaluated approximately 900 files across three competitive categories with 10 voting members, revealed several moderate improvements in voting behavior. The BARS implementation generated more centered ratings, substantially reducing skew in board-level votes and minimizing the "ceiling effect" that is prevalent during similar boards. Board member agreement also improved with fewer aberrant votes, meaning that board members could spend less time voting files. BARS also reduced board member reliance on Officer Evaluation Reports as a feature of their decision making by nearly half.

While the BARS methodology successfully created a strengths-based evaluation process, it did not expand the use of the rating scale as anticipated, except for the Colonel (O-6) population. These findings suggest that BARS offers a promising framework for enhancing centralized selection board processes, though further refinements are needed to address the tendency toward overly centralized score distributions.

57 sec read

DAC develops analytic tools and conducts system-level analysis, to include lethality and vulnerability associated with direct fire weapons. One such tool, the Integrated Multi-burst Probabilities for Accuracy and Kill Tool (IMPAKT3), was developed by DAC to provide estimates of hit probability and probability of kill to determine weapon system effectiveness for direct fire munitions under various engagement conditions. IMPAKT3 calculates these probabilities for direct fire threats against vehicle targets from a single shot up to three multi-round bursts. The tool takes delivery accuracy data, a target vulnerability file, and engagement information as inputs. Inputs may be given in three ways: the analyst may directly manipulate the inputs within the GUI, test data may be loaded and analyzed automatically within the tool using linear mixed models to calculate the delivery accuracy parameters, and inputs may be listed in an input file using batch mode. When loading in test data, confidence intervals may be calculated using parametric bootstrapping. Unlike other direct fire accuracy and kill applications, IMPAKT3 does not calculate the probabilities through Monte Carlo simulations. Instead, it calculates hit probabilities and probabilities of kill through numerical integration delivering stable results quickly.

1 min 50 sec read

The Center for Strategic Competition (CSC) Strategic Opportunities, Awareness, and Risk (SOAR) Dashboard is a framework and tool that has evolved from original investments by the CSC in the Power and Resiliency Index and Strategic Awareness Dashboards that provide a capability to improve strategic awareness. Strategic awareness is defined as the acquisition, organization, and analysis of relevant data essential for decision-making at a geopolitical level. Successful outcomes include data-driven identification of unexpected trends (or confirmation of anticipated trends) and data-driven context for operational and tactical planning. SOAR enables analysts from across the Defense department and the Intelligence Community to quantify, identify, and visualize country and regional standings, trends, and longer-term patterns in global strategic competition.

The methodology behind SOAR provides a systematic approach for visualizing bilateral metrics of influence and relations between hegemonic powers (e.g. U.S., China, Russia) and countries of interest as well as visualizing within-country trends and outcomes. The implicit model of influence driving this framework is based on three assumptions, (1) a goal of the U.S. is to build and strengthen relationships with countries so they align their policies with economic and national security objectives of the rules-based order, (2) economic alignment with a global power or within-country fragilities may influence the decision-making of a country's leadership to align politically with one or another power, (3) the strength of a country's internal institutions may correlate with its degree of autonomy and resistance to economic or political pressure from authoritarian powers. The SOAR Tool enables simultaneous cross-country comparisons along multiple different metrics that visualize changes in these countries over the last 15 years. The data used by the tool spans dozens of sources and over 100 data streams.

This tool has generated country specific deep dives and provided our sponsors with a framework for building strategic awareness and decision making. For example, SOAR dashboard was used to develop country level reports in the early days of the Ukraine war and for regional analysis by analysts within combatant commands. The dashboard is flexible and can be adapted to new use cases in less than a month.

©2024 The MITRE Corporation. ALL RIGHTS RESERVED. Approved for Public Release; Distribution Unlimited. Public Release Case Number 24-3309

57 sec read

This abstract is part of a broader series of abstracts from U.S. Army Test and Evaluation Command (ATEC), where we will discuss Digital Transformation efforts underway in ATEC. ATEC's senior leaders have challenged our workforce to improve the efficiency and efficacy of the test and evaluation (T&E) portion of the acquisitions process while significantly reducing live testing by 2030. Various efforts currently underway provide the components necessary to achieve this vision.

Central to this transformation is the ATEC Data Mesh (ATECDM), a cloud-based infrastructure that supports a data-centric approach to test and evaluation. The ATECDM leverages Azure-based cloud services to ensure data is Visible, Accessible, Understandable, Linked, Trusted, Interoperable, and Scalable (VAULTIS).

Key components of the ATECDM include Hermes, a data upload and access utility that automates metadata tagging and integrates with data governance functions, and the Common Reference Model (CRM), which provides a single access point to authoritative data sources, standardizing data analysis and reporting across the ATEC enterprise. Together, these capabilities enable continuous evaluation of Army systems, enhancing the efficiency and effectiveness of the T&E process and aligning with ATEC's broader digital transformation goals.

1 min 3 sec read

Military readiness is a cornerstone of ensuring the capability of the DoD to fight and win the nation’s wars. The personnel readiness subset is critical to ensuring the Army has the required force structure to deploy when necessary and maintain deployments using the Regionally Aligned Readiness and Modernization Model unit life cycle. There are many components to personnel readiness, but at the core, it comes down to whether a unit has the appropriate number and composition of personnel to accomplish its mission. With an all-volunteer force, there are various reasons why Soldiers choose to continue serving or transition out of the military. We can use system dynamics modeling to best inform senior leaders on the future impacts on personnel readiness.

This research uses the enlisted personnel who make up an Armored Brigade Combat Team as a case study to build out the model. The US Army annually conducts the Department of the Army Career Engagement Survey, highlighting the top ten reasons a Soldier remains in or leaves the Army. This information is combined with policy levers and deployment decisions and then incorporated into a System Dynamics model using Vensim software that can be used as senior leaders make decisions in the DoD and show the impacts on personnel readiness.

51 sec read

Abstract: Effective Maritime Domain Awareness (MDA) is crucial for comprehensive situational awareness in the Indo-Pacific theater, spanning air, land, maritime, space, and cyberspace domains. For the 1st Multi-Domain Task Force (MDTF), MDA is a critical component of its mission, enabling effective joint operations and information sharing with multinational partners. This project details the development of an advanced Maritime Domain Awareness Common Operating Picture (MDA COP), a web-based solution engineered to meet these complex challenges, surpassing the limitations of traditional business intelligence tools like Power BI. Built with a modern React and Vite frontend, the MDA COP delivers a highly responsive, interactive, and user-centric platform. This approach enables enhanced real-time data fusion and visualization, sophisticated geospatial alert management, and tailored user experience. The MDA COP provides a significant leap forward from existing systems, offering superior interactivity and real-time data fusion specifically tailored to the dynamic and collaborative nature of MDTF operations in a contested, multi-domain environment, ultimately optimizing real-time situational awareness and rapid response through a flexible and performant user interface.

1 min 35 sec read

Abstract: Headquarters, Department of the Army G-8 and Office of the Secretary of Defense, Cost Assessment and Program Evaluation commissioned an analysis of alternatives (AoA) to evaluate the performance, schedule risk, cost, and operational effectiveness for a range of XM30 alternatives. The XM30 AoA informed system capability requirements and acquisition decisions on possible replacements for the M2 Bradley Fighting Vehicle. The analysis team assessed different alternative vehicles’ abilities to address Army capability gaps for offensive, defensive, and enabling operations within combined arms maneuver supporting large-scale combat operations. The conclusions of the AoA provided context and analytical support to a milestone B decision and supported the refinement of the XM30 Capability Development Document.

Operational effectiveness analysis considered the impacts of six independent capability trades: sensor, Unmanned Aerial System, Commander’s Independent Weapon System, anti-tank guided missile, Vehicle Protection Systems, and main gun caliber across two independent operational scenarios. Traditional point-solution analytic approaches can obscure the individual impact of each trade due to confounding factors and non-orthogonality of responses. This presentation will cover the design of experiments approach used to explore the six capability areas, with a focus on targeting individual capability trades. The design of experiment (DOE) methodology systematically varied the levels of each trade and statistically analyzed the resulting system operational effectiveness. The DOE enabled the identification of main effects as well as interactions allowing for a deeper understanding of how each trade individually contributes to the overall system response, facilitating informed decisions and maximizing the explorable design space.

Additionally, the presentation will highlight the post-processing techniques and interactions with warfighters and other subject matter experts utilized for operational effectiveness analysis to directly address analytic questions. Finally, the presentation will address how the analysis team created both defensible visualizations and summarized output data that clearly and efficiently conveyed key information to stakeholders to enable informed decisions, ending with a discussion on lessons learned.

1 min 3 sec read

Abstract: Simulation integration is time and resource intensive due to a largely manual process supplemented by isolated scripted tools, often requiring analysts to spend weeks reconciling data inconsistencies and reformatting inputs, while the current research and analysis environment demands shorter project timelines. Rapid advances in AI tools show potential for automating complex workflows, such as data preprocessing, force structure alignment, and scenario generation. However, they have not been tested in integration use-cases, leaving their broader applicability and development requirements uncertain. Developing a dynamic understanding of the opportunities, limitations, and resource requirements for utilizing AI tools for simulation integration is a critical step to inform modernization efforts.

 This project developed a proof-of-concept AI tool for the subtask of importing force structures. Our proof-of-concept tool uses an agentic workflow to ingest a variety of modalities such as images of unit icon wire diagrams, determine the unit characteristics and relationships, then translate the information into simulation entities. This process demonstrates potential for a tangible change to the integration workflow by decreasing the time required for manual data handling. Additionally, this presentation will examine lessons learned about current AI tool capabilities as they relate to the integration process, best practices for AI tool design, and considerations for future AI tool development/deployment.

29 sec read

To contribute to the growing field of predictive maintenance, we created a simulation to model a fleet of vehicles with a particular focus on how parts are ordered for that fleet. The simulation builds on Warren Powell's work in modeling sequential decision problems. Using this simulation, we compared how different ordering policies may optimize a fleet's operational readiness, then explored the effect of artificial intelligence when integrated into these ordering policies. The goal was to produce a framework that provides insight into how AI can improve current practices and help leaders more effectively utilize AI in their units.

1 min 41 sec read

Introduction: Multi-omics data integration plays a pivotal role in extracting meaningful insights from complex biological systems. We developed an R-based data analysis pipeline tailored for multi-omics integration and analysis, using mass spectrometry data as its core analytical output. The pipeline was applied to a multi-organ-on-a-chip model consisting of heart, skin, and lung tissues exposed to various toxic chemicals. This approach harmonizes data from proteomics, metabolomics, and lipidomics to reveal critical features and correlate them to biological pathways and potential therapeutic targets. The implications of this work include using the results to develop broad medical countermeasures to toxins.

Methods: Data from a multi organ-on-a-chip study was first imported into R and separated by organ and chemical exposure type. Numeric and categorical identifiers were associated with each feature to simplify downstream processing. Data from proteomics, metabolomics and lipids were combined into one data table and subject to a PCA and PLSDA analysis. The most important features identified in the PLSDA were extracted and input into MetaboAnalyst for pathway mapping. The biomolecules involved in high impact pathways were further investigate by comparing relative abundance to identify critical biomarkers relevant for medical countermeasure development.

Preliminary Data: The pipeline successfully integrated multi-omics datasets from a multi-organ chip study where heart, skin, and lung tissues were exposed to toxins for 24 hours. Initial PCA revealed clusters of data points corresponding to the 3 organs and toxins. Subsequent PLS-DA identified key metabolites, lipids, and proteins driving the variance, enabling mapping of disrupted metabolic pathways. For example, exposure to a chemical stressor resulted in significant perturbation of energy and amino acid metabolism pathways in cardiac tissues, while skin and lung tissues displayed unique markers of stress and inflammation. The pipeline's automated feature labeling and pathway mapping provided an efficient means of generating biologically meaningful insights. Preliminary results suggest its utility in identifying critical biomarkers and therapeutic targets for chemical countermeasure development.

Novel Aspect: The Automated R pipeline integrates multi-omics data to identify treatment-specific pathways in multi-organ-on-a-chip studies. Conflict of Interest: No conflicts of interest.

1 min 48 sec read

Government agencies to explore, visualize, and communicate with data, the Army continues to primarily rely on PowerPoint and Excel for this purpose. Efforts to leverage other tools such as R Shiny, Tableau, and Python's Dash, have offered some success, but are often hampered by issues such as licensing, database repository accessibility, and network compatibility and security issues with server-based applications. This is especially challenging when attempting to share results across multiple organizations which may operate on different network configurations or are unable to use products other than Microsoft-based Army 365 services.

Bokeh is a powerful tool for developing highly interactive, web browser-based visualizations that allow users to explore data in real-time, making it ideal for analyzing complex data sets which are completely self-contained and can be shared within a single Hypertext Markup Language file. This presentation explores interactive plotting capabilities of the Bokeh Python library, with a primary focus on how The Research and Analysis Center (TRAC) analysts applied it to transform complex datasets into dynamic and responsive dashboards that have been used to underpin recent major analysis efforts. It also introduces Bokeh's core features, emphasizing its ability to create plots that respond to user interactions such as zooming, panning, linking, and hovering which enables analysts to rapidly interpret data.

The presentation will first focus on constructing interactive dashboards that have been tailored to the needs of Army simulation post-processing by demonstrating how Bokeh enables the creation of visualization components and show how these can be combined into cohesive, user-friendly dashboards, where users can explore their data by integrating widgets such as sliders, drop-down menus, and buttons on the fly to create a detailed and customized view of the results. This interactivity aids in highlighting patterns, trends, anomalies, and defensible responses to Army Senior Leader questions which are essential for informing decision-making. The presentation will also provide an overview of transforming raw data into the appropriate format, including handling time-series data, performing aggregation, and reshaping data structures to work within Bokeh. Finally, we will discuss some of TRAC's future innovation efforts for more data-immersive presentations and summarize the lessons learned from utilizing this tool over the past year.

1 min 16 sec read

This research optimizes the number, placement, and capacity of Military Entrance Processing Station (MEPS) facilities in the United States to minimize applicant and recruiter transportation requirements to improve operational performance and increase recruitment efficiency. Two models are developed using mixed-integer programming to address major constraints such as facility capacity, budget limitations, and geographic coverage: a capacitated facility location problem (CFLP) and a maximal covering location problem (MCLP). These models utilize applicant densities, recruiter locations, existing MEPS sites, and candidate MEPS cities to drive a data-driven, scalable optimization framework. Extensive computational testing, sensitivity analysis, and scenario-based evaluations highlight significant opportunities to reduce applicant travel distances and mitigate capacity imbalances, demonstrating clear advantages over current practices. For instance, the solution to a monolithic CFLP model adds three new MEPS to the network, providing significant operational gains: the total annual applicant travel distances decrease by 1.2 million miles in Florida, 1.2 million miles in Texas, and 1.0 million miles in California. The CFLP model also enhances applicants' accessibility within 60 miles of a MEPS, with an increase of 18.4% of applicants in Florida, 5.2% in California, and 4.0% in Texas, while also relieving the problem of overutilization at high-demand facilities. This data-driven approach equips the United States Military Entrance Processing Command (USMEPCOM) with a flexible decision-support tool to optimize the MEPS network design. These findings balance the reduction of operational costs with applicant accessibility in actionable insights for strategic facility planning, supporting the USMEPCOM mission to sustain a ready and capable military force by ensuring the most efficient methods of meeting accession requirements.

46 sec read

The Discrete Event System Specification (DEVS) software framework, with its robust and flexible structure, has shown potential for advancing Modeling Simulation as a Service (MSaaS). By leveraging DEVS, software architects can design simulation environments to be highly modular and interoperable, allowing for more rapid composition and deployment of simulation services in cloud-based infrastructures. This capability is crucial for MSaaS, since it allows for the creation of dynamic, simulation environments tailored to user needs. The use of DEVS with MSaaS promises to enhance the scalability, efficiency, accessibility, and maintenance of simulation services, making them more cost-effective and user-friendly. Future research should include optimizing DEVS for cloud environments, enhancing MSaaS interoperability standards, and exploring DEVS application within the Army's doctrinal directive to conduct modeling and simulation in a multi-domain operations paradigm. This presentation will provide an overview of DEVS, MSaaS, and their implications to the ongoing Center for Army Analysis campaign model modernization effort.

1 min 33 sec read

The Warfighter can make use of a wide array of Positioning, Navigation, and Timing (PNT) sensors to estimate the location of a target of interest. The metric used to assess and compare a sensor's capability to estimate a target location is known as Target Location Error (TLE). The Department of Defense, along with the private sector, propose a wide array of PNT sensors to meet this need. The U.S. Army Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) developed a model that estimates the performance of these novel PNT sensors even if the exact algorithm that the sensor employs is proprietary or unknown. The TLE Calculator Tool can calculate the TLE and/or timing error for multiple sensor types including Electro-optical/Infrared (EO/IR) sensors with Laser Range Finders (LRF), and Signals Intelligence (SIGINT) sensors. The tool was recently enhanced so it can model how the random noise from the imaging sensor of Laser Warning Receivers (LWR's) can affect the sensor's TLE. LWR's measure the direction of incoming laser radiation from sources such as: LRF, laser target designator munitions, and laser beam riders. In situations where the intensity of the incoming laser radiation is low, the random noise on the imaging sensor can have a large effect on the final TLE of the sensor. It is therefore critical to understand how this noise can affect the final location measurement of an LWR. This technique can also be expanded to model other sensors that use imaging arrays or whose location measurement is dependent on a collection of sensors that have potentially large amounts of random noise. This presentation will contain an overview of the TLE Calculator Tool, an overview of the Noise Model enhancement, steps taken to validate the enhancement, and sample calculations comparing TLE results for the baseline model with results that consider the addition of the Noise Model.

1 min 36 sec read

This abstract is part of a broader series of abstracts from the U.S. Army Test and Evaluation Command (ATEC), where we will discuss digital transformation efforts underway in ATEC. ATEC's senior leaders have challenged our workforce to improve the efficiency and efficacy of the test and evaluation portion of the acquisitions process while reducing live testing by 30% by 2030. Various efforts currently underway provide the components necessary to achieve this vision. Testing and evaluating interoperability and effectiveness of systems in a Multi-Domain Operating (MDO) environment presents a unique challenge for the Army test community. MDO is the Army's war fighting concept that involves the integration of all domains of warfare as part of the joint force. For example, how does a system under the test perform as part of a division level system of systems operation? What are the effects of cyber and electronic warfare threats in an MDO environment? Are they localized, or do they have broader implications? To meet this challenge, ATEC developed and demonstrated an MDO scale distributed live, virtual, and constructive (LVC) test environment, which connected every ATEC test site and the subordinate organizations from across the country. In an initial operational capability demonstration, more than 60 systems, simulations, and data collection elements interacted cooperatively to execute seven land, air, space, and cyberspace scenarios. Real-time data collection, display, and cloud-based data analytics were employed to assess the effectiveness of the network as well as all the connected live and virtual systems. The environment was underpinned by an innovative closed restricted test network that can support security levels ranging from unclassified to top secret. As more unmanned systems, sensors, and weapons are integrated into the formation and networks are used to enable rapid decision making, the need to test at scale becomes increasingly apparent. The most practical and cost-effective solution is through the use of LVC simulation. This briefing will cover why and how distributed LVC is the solution.

45 sec read

The Discrete Event System Specification (DEVS) software framework, with its robust and flexible structure, has shown potential for advancing Modeling Simulation as a Service (MSaaS). By leveraging DEVS, software architects can design simulation environments to be highly modular and interoperable, allowing for more rapid composition and deployment of simulation services in cloud-based infrastructures. This capability is crucial for MSaaS, since it allows for the creation of dynamic, simulation environments tailored to user needs.

The use of DEVS with MSaaS promises to enhance the scalability, efficiency, accessibility, and maintenance of simulation services, making them more cost-effective and user-friendly. Future research should include optimizing DEVS for cloud environments, enhancing MSaaS interoperability standards, and exploring DEVS application within the Army’s doctrinal directive to conduct modeling and simulation in a multi-domain operations paradigm. This presentation will provide an overview of DEVS, MSaaS, and their implications to the ongoing Center for Army Analysis campaign model modernization effort.

1 min 52 sec read

Abstract: While establishing Joint Task Force Southern Border (JTF-SB), the 10th Mountain Division (10MTN) operationalized existing data platforms to enhance situational awareness, decision-making, and staff analytic capacity. 10MTN exploited the immature theater and seized the opportunity to implement core tenets of the Army Data Plan by establishing effective data governance, structure, and interoperability in a multinational, joint, and interagency environment. This implementation enabled the creation of a multi-domain common operating picture, facilitated seamless information flow from the forward line of troops – such as a Squad Leader or Border Patrol Agent – to the combatant commander, and unlocked the potential for staff analysis and the application of machine learning and other artificial intelligence tools.

To achieve this, 10MTN employed an iterative methodology focused on the structured collection and analysis of critical information requirements. We prioritized capturing standardized reports, Commander’s Critical Information Requirements (CCIRs), and Priority Information Requirements (PIRs) in a structured, machine-readable format. Capturing these reports involved developing standardized data templates and automated ingestion pipelines to integrate information from diverse sources. Subsequently, we integrated unstructured data such as operational graphics and meeting transcripts to create a holistic, digital representation of our environment to facilitate analysis. Importantly, this transformation was achieved utilizing only assigned personnel, existing data platforms, and a culture of innovation and resourcefulness.

This structured data collection enabled the rapid development of a unified Common Operating Picture (COP) accessible to decision-makers at all echelons, which is crucial for maintaining situational awareness across domains. Within 100 days of assuming the mission and following just six months of training with the provided data platform, 10MTN began to apply large language models and other predictive analytic techniques to the collected data. These models and techniques informed data-driven tactical and operational decisions regarding force posture, justification for additional authorities, and optimization of patrol times and locations. The resulting data structure and tools significantly improved operational data accessibility for leaders and staff at all echelons, enabling them to query and analyze the data.

The success of this approach stemmed from a focus on operational requirements, continuous improvement, and maximizing platform capabilities. By prioritizing data standardization and automated analysis, 10MTN transformed raw information into actionable intelligence, enhanced situational awareness, and accelerated decision-making in multi-domain operations.

1 min 21 sec read

Abstract: Effective warfighting and force employment hinge on informed decision-making, often within compressed timeframes. While the military excels at analyzing threat networks, understanding our own organizations is crucial for optimizing operations, training, and strategic partnerships to maintain lethality especially during times of organizational change. This presentation highlights how the U.S. Army Training and Doctrine Command (TRADOC) G-2 employed network science to map its internal and external engagements over a six-month period. An interview-based survey of 29 individuals representing 26 teams was used to produce internal and external networks across several organizational dimensions including work interactions, collaboration on products, delivery of products, decision making, formal reporting to, and social closeness. Findings from a battery of network science measures and visualizations suggest that internally TRADOC G-2 is a ‘small world’ with teams maintaining a high-level of internal cohesion (In-Degree Centrality M=12.19, SD=4.06). Externally, TRADOC G-2 was found to have 587 ties to 315 organizations over a six-month period with variation between ‘local-bonding’ teams without many external connections and teams and with more ‘cosmopolitan-bridging’ teams with many interactions outside the organization (Out-Degree Centrality Mean=22.58, SD =16.11). This analysis allowed greater understanding of how TRADOC G-2 achieves its mission after a change in leadership and organizational restructuring. This analysis resulted in leadership having a detailed map of their organization beyond the formal organizational chart and led to several improvements for continued analysis through an automated tracking system. Additional study is needed to understand if TRADOC G-2's organizational structure is common or unique among other Army organizations and how the agent-based modeling of these dynamics may show opportunities for deliberate transformation that positively impact Army productivity and well-being.

1 min 15 sec read

Abstract: The processing and analysis of qualitative data is a labor-intensive task for researchers in Operations Research and other fields. Qualitative data such as text, audio, and images may be collected from interviews, surveys, workshops, and various other data collection methods. Significant amounts of qualitative data are also present in document repositories or knowledge bases. This data is primarily unstructured or semi-structured, yet it contains valuable information that needs to be extracted, structured and analyzed to gain insights. Recent advancements in AI technologies have the potential to support and automate qualitative content analysis. These advancements can revolutionize the field by enabling researchers to efficiently analyze large volumes of qualitative data, uncovering insights that were previously inaccessible due to resource constraints.

The Bundeswehr Office for Defense Planning is currently exploring the application of AI methods for qualitative content analysis in the military context. To this end, 13 use cases have been identified and specified, and an AI platform has been built to provide an AI toolkit for this research purposes. Throughout the research project, AI-driven data processing pipelines will be developed to experimentally assess the effective application of AI methods to these use cases. Additionally, aspects concerning the provision, operation, and maintenance of AI services within a highly secure on-premise IT environment will be investigated further.

This presentation will outline the most relevant AI use cases identified, present initial findings regarding the effective application of AI methods to these use cases and provide guidelines for sustainable AI service operation.

40 sec read

With the goal of increasing the operational readiness in U.S. Army Aviation, we seek to employ machine learning methods that account for censored data in predictive maintenance models. Army maintenance records for aircraft form large datasets with significant proportions of censored data, meaning they are records for parts that have not yet failed. Traditional models often ignore this valuable information, leading to biased or incomplete predictions. Using a survival analysis framework to model predictive maintenance, we primarily explore the Buckley-James estimator combined with penalized regression methods to better handle censored data using UH-60 Black Hawk maintenance records. In collaboration with the Artificial Intelligence Integration Center's (AI2C), we aim to further integrate blended models accounting for censored data to equip commanders and maintainers with enhanced tools for proactive decision making and increased fleet readiness.

56 sec read

The Operations Research Systems Analysis Military Application Course (ORSA-MAC) is a critical pathway for developing FA49 officers capable of addressing complex DoD challenges; however, a historical attrition rate of approximately 13% results in a significant loss of training investment – averaging $54,780 and 1,760 operational hours annually. To mitigate this, the ORSA Community at Army Sustainment University (ASU) implemented a standardized Fundamental Skills Exam (FSE) for all ORSA-MAC attendees. The FSE assesses baseline proficiency in algebra, probability, and calculus – skills essential for success in the ORSA-MAC curriculum. The FSE serves as a diagnostic tool, identifying areas where students may benefit from targeted support, and proactively addressing potential academic challenges. Consequently, these results are leveraged to inform alterations to student onboarding procedures and course material, fostering a continuous cycle of improvement. This data-driven approach facilitates proactive engagement with students, enhancing their preparation and providing targeted support. Students achieving 80% or above demonstrate a high probability of course completion, reflecting a strong correlation between FSE scores and graduation rates. These findings underscore the FSE's value in identifying students who may require additional preparation. The FSE ultimately contributes to a stronger cohort of FA49 analysts.

1 min 8 sec read

The Army is actively developing modernized, advanced powertrains for ground vehicles to make Warfighters more lethal and enable longer operations. However, because these powertrains offer capabilities beyond existing and legacy systems, ground-vehicle representations in combat simulations must be updated accordingly.

In response, the DEVCOM Analysis Center led the development of a new Physical Knowledge Acquisition Document (PKAD) to integrate advanced powertrains into combat simulations. This effort picks up where the Fuel Consumption Prediction Model (FCPM)-based PKAD work left off, adding improved vehicle-energy-management algorithms and other enhancements in the MATLAB development environment. In addition, the accuracy of this system-level approach—tailored to match the fidelity of combat simulations—is being compared with higher-fidelity item-level modeling approaches and authoritative test sources. The MATLAB environment has allowed for more-modular tool design and rapid iteration with both test data and simulated component data.

The resulting data products are intended to integrate with standard file formats used in combat simulations, but the same approach can also support acquisition studies and sustainment analyses. Future studies will require modeling that keeps pace with rapidly diversifying vehicle architectures, especially robotic or autonomous platforms that may rely on unconventional power and propulsion technologies. This approach reduces turnaround time for system performance data requests and provides a path for rapidly integrating emerging propulsion technologies into analyses to inform the decision trade space

1 min 19 sec read

Processing data at the edge is critical for the Army to maintain decision dominance over its adversaries. The Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) conducts analysis of the performance and cyber resiliency of Army systems and is developing methods and metrics to assess data processing approaches. In the conduct of these efforts, DAC found that while Federated learning (FL) models can process large amounts of heterogeneous data, maintain data security and keep the communication (bandwidth usage) costs low, they perform poorly when the class (traffic) distribution for the clients is locally and globally imbalanced. Data collected from Army systems is almost never a class balanced dataset.

Over the past few years, DAC has led a series of efforts to explore the performance impacts of various frameworks building upon the original work, including Multi-Model-based Federated Learning (MMFL) and a novel client data augmentation strategy. The most recent effort, and the focus of this paper, is on updates that increase cyber resilience and improved performance. This presentations will cover: (1) An overview of the updated MMFL framework to analyze imbalanced datasets and dynamically group clients to account for clients leaving / joining the network; (2) A data secure Diffusion model based MMFL (D-MMFL) to introduce unknown-attack traffic to each client to enable analyses of cyber resilience of distributed networks in fields of operation; (3) A synthetic dataset that includes local class and global data imbalance issues used to conduct the analysis / assessment; and (4) Performance results of the D-MMFL using the commonly used traditional and the new synthetic datasets to demonstrate the robustness of the proposed framework(s).

43 sec read

The US Army needs the ability to analyze world news feeds at a scale to support information
advantage requirements. Past efforts have focused on acquiring, parsing, enriching, and
loading large open-source news feeds that include Common Crawl News and the Global
Database of Events, Language, and Tone. This filtered and enriched data is now being
ingested on the Information Advantage instantiation of the Big Data Platform, which is
commonly referred to as “LOKI.” These automated efforts push approximately 500K articles
per day to LOKI. Our current effort details the development and deployment of dashboards in
Apache Superset to help analysts understand the narratives, networks, and geography
represented in large open-source news datasets. We will discuss the power of SQL data
manipulation through the Ionic Data Application and then the unique visualizations that we’ve
used to understand emerging events and actors through these large datasets.

51 sec read

In support of Large-Scale Mobilization Operations (LSMO) and sustained Large-Scale Combat Operations (LSCO), this presentation introduces a simulation-driven framework to model the (Re)constitution of Army units during Phase IV operations. Focused on transforming civilians, retirees, and Individual Ready Reserve (IRR) personnel into Military Occupational Specialty (MOS)-qualified Soldiers, the study integrates historical data from the
Selective Service System (SSS), Military Entrance Processing Command (MEPCOM), and U.S. Army Training and Doctrine Command (TRADOC).

Using ProModel and SimPy, we develop scalable, discrete-event simulations that map the end-to-end mobilization pipeline
across MEPS, training facilities, and Force Projection Training Sites (FPTSs). These tools enable planners to test key variables—such as arrival rates and training capacities—and identify system bottlenecks under various mobilization conditions. Housed in the Army Vantage environment, this simulation capability supports real-time analysis and scenario planning, empowering strategic decision-makers to optimize resource allocation and enhance readiness. The findings underscore the need for synchronized efforts among SSS,
MEPCOM, TRADOC, and operational commands to ensure rapid and efficient force generation for protracted conflict.

39 sec read

Abstract: The SOUTHCOM J86 Operations Research Division is pioneering a novel approach to extracting actionable insights from unstructured text data by leveraging Large Language Models (LLMs) and the "Text as data" framework. Through the identification, consolidation, and analysis of large unstructured datasets using LLMs, this study demonstrates the effectiveness of Natural Language Processing in uncovering key information, including themes, topics, and named entities. The application of LLMs to Natural Language Processing tasks enables the analysis of complex texts, such as combatant command posture statements, revealing subtle shifts in priorities and thematic patterns. Ultimately, the findings of this study underscore the potential of this approach to inform decision-making and strategic planning, with significant implications for refining plans, operations, and exercises, as well as identifying trends and patterns in complex textual data.

1 min 1 sec read

Abstract: Cost analysis is essential to the Department of Defense acquisition process, offering insights into budgets, cost drivers, and supporting informed strategic decisions. Power BI enhances this by integrating data from multiple sources—Excel, databases, and cloud services—into a unified dataset. This consolidation simplifies complex data into clear, tailored reports for each project.

Power BI’s powerful visualization tools create interactive reports that break down systems into subsystems, highlight cost drivers, and reveal emerging insights. Stakeholders can drill down into the data, filter results, and perform ad-hoc analyses. For example, in a recent Counter-Small Unmanned Aerial System study, Power BI helped break down near-term systems into components to forecast mid-to-long-term costs and trades. These visual aids quickly highlight trends, anomalies, and areas for further investigation.

With real-time analytics, Power BI enables up-to-date cost analysis, essential for short study timelines. Its collaborative features foster transparency by sharing insights across stakeholders, ensuring alignment with the same data. This approach not only streamlines analysis but also empowers decision makers with timely, accurate information.

This presentation will demonstrate how to implement Power BI for cost analysis, covering data linking, transformation, and the creation of collaborative dashboards for effective decision-making.

44 sec read

Abstract: Alternative vehicle propulsion technologies, whether taken alone or in conjunction with internal combustion engines, offer many compelling advantages to the Warfighter, such as silent operation, increased range, fuel savings, and power export capabilities. While these technologies have seen continued maturation in the commercial ecosystem, it remains to be determined whether or not they are sufficiently robust, economical, and effective for integration into Army and Marine Corps systems. To that end, DEVCOM Analysis Center has conducted a business case analysis study with the goal of providing insight into the costs and benefits of implementing alternative vehicle propulsion technologies across multiple platforms and at several scales.

This paper will discuss our methodology, as well as the development thereof, for characterizing the reliability change associated with various levels of drivetrain modernization, as well as supporting the larger business case analysis effort in a data-sparse environment.

1 min 28 sec read

Inconsistencies in how battalion maintenance teams prioritize and allocate mechanics to planned maintenance and repair tasks can have significant impacts to readiness for the 82D and the Immediate Response Force. Although all 50+ battalions rely on similar principles—fixing the most important equipment first—the business rules they apply vary, and outcomes are uneven. Meanwhile, fluctuations in mechanic availability due to appointments, training, and leave make daily planning complex and often reactive. To address this, we are developing a two-phase model to bring analytic rigor and predictive insight into battalion-level maintenance planning. Ultimately the battalions' maintenance officer and tech will have an automated tool that delivers an optimal solution from which they can adjust fire. In Phase One, we are building an optimization tool that allows maintenance techs to input available personnel by day and skill level, then outputs an assignment plan that maximizes readiness by prioritizing PACER and mission-essential systems, while accounting for part availability, qualifications, and shift constraints. The model leverages GCSS data, including repair time estimates, MOS requirements, and part arrival timelines, to generate realistic work plans that balance readiness and throughput without overemphasizing mechanic utilization. In Phase Two, we shift from planning to forecasting. Our goal is to predict job-level maintenance demands up to four weeks in advance, using trends in fault generation and historical repair patterns. This will allow battalions to anticipate upcoming workload surges, assess whether available man-hours are sufficient, and make deliberate resource decisions—such as adjusting shifts, combining teams, or planning anticipatory maintenance—to avoid degraded readiness. This project is designed to be actionable at the battalion level, repeatable across formations, and aligned with operational priorities. Our aim is to give maintainers the tools they need to make better decisions, faster—and ultimately keep critical equipment in the fight.

35 sec read

While traditional acquisition systems have established methods for mitigating hardware, software, and human-systems integration failures, the rapid adoption of artificial intelligence (AI) presents unique reliability challenges. DAC recently responded by developing the AI Failure Mode Wheel, cataloging 30 AI-specific risks. Building on this foundation, DAC has introduced the AI Reliability Scorecard – leveraging the well-established Reliability Scorecard methodology – to systematically evaluate an AI program's preparation and execution of critical activities during design, test, and deployment. This presentation will detail the DAC AI Reliability Scorecard and describe how its outputs highlight critical risk areas that link directly to potential failure modes identified in the AI Failure Mode Wheel, ultimately informing targeted resource allocation and enabling improved mission reliability.

43 sec read

Building upon the success of crowdsourcing initiatives in defense, Git Your Tactics Straight: Crowdsourcing 2.0 dives deep into our ambitious endeavor to revolutionize mobilization through collaborative innovation. This presentation moves beyond the initial call to action, establishing interlocking fields of fire for data collection and analysis. We will showcase progress made in leveraging crowdsourced insights to achieve a comprehensive understanding of Army mobilizations past, present, and future.

Discover the tangible achievements of the past year, including the breakdown of data silos, accelerated decision cycles, and identification of key capability gaps. Looking ahead, we'll delve into the transformative potential of human-machine teaming, AI systems trained on military jargon and logic, all powered by a unified data platform. Join us to explore how we are harnessing the power of the crowd to build a more agile, informed, and effectively mobilized fighting force for the future.

45 sec read

As part of modernization efforts within US Army technologies, many systems are being updated or designed to utilize artificial intelligence (AI) or other forms of autonomy. The Combat Capabilities Development Command (DEVCOM) Analysis Center (DAC) is conducting reliability analyses on these systems to assess their potential effectiveness. These new autonomous elements must be factored into traditional reliability models and calculations to ensure that the analyses conducted are as accurate as possible. System-theoretic process analysis (STPA) is a form of system-level analysis that uses a control structure to identify unsafe control actions and (subsequently) loss scenarios. This analysis can be used to construct a Bayesian Belief Network, which in turn can determine conditional probabilities for failure states within the system. STPA has potential to be very useful in assessing system reliability, and this paper seeks to examine its ability to model threats to AI elements within a larger system.

47 sec read

The Joint Integrated Contingency Model (JICM) is an extremely complex, theater-level model that encompasses the strategic and operational levels of land, air, and naval warfare with a global set of models and databases. The Center for Army Analysis (CAA) owns the model and leverages it to conduct campaign analysis of operations plans and Defense Planning Scenarios. This year, CAA overhauled its JICM training program, which was historically taught by RAND as a two-week in-person course. The new training program is comprised of four modules: LINUX Familiarization, JICM Basics and Tutorial Exercises, Advanced Practical Exercises, and a Capstone Project. The program consists of over seventy-five video-based presentations and tutorials, enabling new analysts to complete the course asynchronously at their own pace. The course is also capable of being exported to external JICM users, such as United States Forces Korea and the Korean Institute for Defense Analyses, thus eliminating the need to send analysts stateside for training.

58 sec read

As data analysts and scientists, we are facing increasing demand to incorporate generative AI to drive insights and inform command decisions. While this enthusiasm is well received, the barrier to entry for designing these workflows has mitigated their practical success. If the operational world is to meet this demand for AI-supported analysis, it is necessary to demonstrate a common framework that can conduct advanced NLP processes in a concise, understandable way. This presentation introduces the concept of Generative AI Logic Blocks, a framework for creating transparent, modular, and reusable components that underpin generative AI models. Through a range of examples, attendees will gain hands-on experience with a system for designing and implementing their own logic block workflow using Foundry Vantage. Additionally, we will discuss common pitfalls and challenges that can arise when working with these generative AI logic blocks and provide practical strategies for mitigating errors and ensuring robust performance. By the end of this presentation, attendees will be equipped with the knowledge and skills to develop their own generative AI workflows, enabling them to unlock the full potential of powerful AI models and drive more effective analysis and decision-making in the field.

56 sec read

Abstract: The U.S. Army Operational Test Command (USAOTC), a component of the Army Test and Evaluation Command (ATEC), plays a vital role in supporting Transformation in Contact (TiC) units as part of the broader Army Transformation Initiative (ATI). USAOTC ensures that TiC units receive precise, actionable test data to inform their decision-making processes and enable effective operational adjustments. To enhance the tracking and management of both historical and future support activities, as well as resource utilization, USAOTC developed an advanced dashboard using Microsoft PowerBI.

This TiC Common Operational Picture (COP) dashboard integrates data from multiple sources, providing real-time visualization and comprehensive analytics that improve situational awareness and support command-level operational decisions. Its architecture employs automated Extract, Transform, Load (ETL) processes to maintain data accuracy and timeliness, enabling detailed performance analysis and efficient resource management. Through interactive visualization tools, users can monitor mission timelines, resource allocation, and key operational metrics within a unified platform. This presentation will showcase the TiC COP dashboard’s design, data integration workflows, and analytic capabilities, highlighting its impact on streamlining support operations and facilitating informed decision-making across TiC units.

36 sec read

Abstract: The Army is working toward reducing the amount of live testing needed and leveraging predictive logistics to inform program decisions, predict critical system failures, impact design improvements, and increase readiness. This requires new analytic techniques as well as additional sources of data. The DEVCOM Analysis Center has developed one such new analytic technique, specifically intended to reduce the maintenance burden and costs for Army systems using advanced machine learning algorithms combined with durability simulator data. This effort uses surrogate vehicle models to replicate Aberdeen Test Center's Heavy Vehicle Durability Simulator (HVDS) simulation data using RecurDyn multibody dynamics software. This briefing describes this new methodology as part of DAC’s initiative to advance machine learning applications to impact readiness and achieve program efficiencies.

43 sec read

Abstract: This paper uses causal methods to estimate the impact of place-based investments in strategic technology industries on local economies. Specifically, we leverage synthetic control methods with economic data from U.S. metropolitan areas to evaluate the effect of Tesla, Inc.’s Gigafactory Nevada – a producer of battery cells and electric vehicle components – on employment, income, and price level outcomes in the Reno metropolitan area via estimating the counterfactual outcomes had the factory not been built. The analysis both provides estimates that can inform the implementation of future strategic technology industrial policies and develops a replicable empirical framework for quantitatively estimating the causal impact of place-based policies.

Copyright 2025 The MITRE Corporation.

Approved for Public Release; Distribution Unlimited. Public Release Case Number 25-1026.

This technical data deliverable was developed using contract funds under Basic Contract No. W56KGU-18-D-0004.

1 min 3 sec read

Functional Data and its Applications to Pressure Wave Curves

Artillery systems generate high intensity pressure waves during each firing event that over time degrades the gun tube, leading to potential misfires and breakdowns. As a part of an ongoing propellant research and development effort, functional data analysis has been used to generate models of these pressure waves to determine the optimal configuration to minimize negative effects on tube lifespan. Functional data analysis is a statistical method that utilizes functional principal component analysis (FPCA). FPCA allows for dimension reduction within a design space that has numerous factors, allowing the analyst to see multifactor interactions. After undergoing dimension reduction, input factors are broken into their principal components that show the degree of influence each factor has on the overall result. The breech, chamber, and differential pressure were analyzed using functional data analysis on live fire and simulation data sets spanning multiple different platforms. Initial results have shown that using this methodology can help shape future live-fire test planning and design of experiments by binning factors of interest to investigate during the event. Future goals for this methodology include increasing the safety, reliability, and gun tube lifespan of artillery systems while simultaneously increasing the cost effectiveness of propellant usage round to round.

1 min 0 sec read

As the military continues to emphasize data-driven decision-making, data literacy training has become a ubiquitous part of analyst development. However, a closer examination of typical data literacy course content reveals a significant gap between the skills being taught and the realities of working with military data. While students are often instructed on statistical concepts like mean, mode, and standard deviation, they are not being equipped with the practical skills to effectively work with the operational data they encounter in their daily jobs. Furthermore, these courses frequently overlook the opportunity to teach analysts how to leverage available tools like Microsoft Excel and data platforms to automate routine tasks, such as counting and chart building. Instead of rushing to statistical concepts that are rarely encountered, we propose a more pragmatic approach: focusing on building a deep understanding of the military's own data and teaching analysts how to extract insights from it. By prioritizing data familiarity and basic analysis skills, we can empower analysts to identify and capitalize on low-hanging fruit within their organizations, driving tangible improvements in operational effectiveness. This presentation will highlight the limitations of current data literacy training and make the case for a more applied, military-context-specific approach to analyst development.

37 sec read

With Artificial Intelligence and Machine Learning (AI/ML) models constantly improving, predictive maintenance is one area where the military has the potential to make significant improvements. Cadets and faculty at the United States Military Academy, in collaboration with the Artificial Intelligence Integration Center (AI2C), are researching applications utilizing currently available data, and running simulations to provide requirements for future AI/ML models. This research effort can be broken down into three lines of effort: evaluating the quality of the current data being utilized to train models, building predictive maintenance models with our current data, and providing requirements for future models utilizing simulations. Utilizing this and future research, we hope to inform more effective use of AI/ML in the predictive maintenance space.

57 sec read

DEVCOM Analysis Center (DAC) is responsible for generating authoritative ground combat system performance data, to include large caliber direct fire weapon accuracy. As the DoD and Army move toward automated systems, understanding AI's impact is essential to properly characterize Probability of Kill given a Shot, and possible errors that may inadvertently arise in future controlled weapon systems.

This presentation will highlight the four key areas of the analytical framework that were developed to achieve this objective. First, the framework addresses the future, new and improved Tank Accuracy Model, which will incorporate wind and Area of Operation specific information. This includes a description of an unmanned aerial system (UAS), or off-board sensor, with the ability to sense cross wind and convey the information back to the home platform. Secondly, the framework touches on the ability of controller designs to accept target location input and move a turret to the centroid of the target in simulations. Third, within the framework, the gunner lay error analysis tasks will be laid out with highlights of key milestones. Finally, the framework will describe the steps for comparative analysis between human and AI technology.

1 min 2 sec read

This abstract is part of a broader series of abstracts from the U.S. Army Test and Evaluation Command (ATEC), where we will discuss digital transformation efforts underway in ATEC. ATEC's senior leaders have challenged our work force to improve the efficiency and efficacy of the test and evaluation portion of the acquisitions process while reducing live testing by 30% by 2030. Various efforts currently underway provide the components necessary to achieve this vision, and ATEC will present the following subset of these efforts:

U.S. Army Evaluation Center Digital Evaluation Tool & Modeling and Simulation Hub.

ATEC Data Mesh, its Data Upload Utility, Hermes, and the Common Reference Model.

ATEC Data Governance.

Distributed Live Virtual Constructive Testing.

ATEC is taking a multi-pronged approach to standardization of the body of test and evaluation data. As a geographically distributed command consisting of subordinate organizations with historical autonomy, myriad silos and differential processes present challenges to transformation of ATEC's digital landscape. This presentation will outline our approach to standing up pilot data governance councils, painting the data landscape, developing recommendations for change, and implementing those changes. The briefing will close with our plans for continuing to expand data governance across the command's disparate mission areas.