Utilizing Large Language Model (LLM)/Generative AI to Develop Energy Calculation Tool for Manufacturing Processes

OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Advanced Computing and Software, Advanced Infrastructure & Advanced Manufacturing, Trusted AI, and Autonomy The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the statement of work (SOW) tasks intended for accomplishment by the FN(s) in accordance with the Announcement. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws. OBJECTIVE: Energy cost has a direct impact on the cost of production. DLA, as a procurement agent for DoD and all of government, understanding the cost elements are essential. Given that DLA procures more than $40 billion worth of items annually, a small percentage reduction in energy consumption can translate to substantial monetary savings. Manufacturing, a cornerstone of modern economies, broadly encompasses two primary domains: discrete and continuous. Discrete manufacturing refers to the production of items such as weapons, military uniforms, etc. In contrast, continuous manufacturing involves non-stop processes producing goods like chemicals, fuels, and certain consumables. Regardless of the category, energy remains a pivotal input, significantly influencing the cost, efficiency, and environmental footprint of the production process. TRL 3. (Analytical and Experimental Critical Function and/or Characteristic Proof of Concept) TRL 6. (System/Subsystem Model or Prototype Demonstration in a Relevant Environment) DESCRIPTION: Central to the discourse on energy efficiency in manufacturing is the concept of value added (VA) and non-value added (NVA) processes. Value-added processes directly contribute to the product's transformation, whereas non-value-added ones, although often essential, don't enhance the product's inherent value from a customer's perspective. By dissecting the energy usage into these two categories and employing precise quantification mechanisms, manufacturers and DoD can achieve a deeper understanding and, subsequently, higher operational efficiencies. This research is based on the following hypotheses: (1) Energy used to add value (product transformation) could be modeled analytically (either physics or chemistry based), (2) Nonvalue added energy would be a function of value-add and it could be derived either from the total energy and value-add energy or allocated using empirical techniques. There are several databases and peer reviewed journal papers are available that shows the total energy required for various manufacturing processes. See References section for examples of where one could get the total energy required for various manufacturing processes. Since the value-added energy in manufacturing is based on physics or chemistry-based equations, one could make use of Large Language Model (LLM)/Generative AI to identify the representative equations, and then use the same LLM/generative AI to develop the python code. Relying on the vast knowledge reservoir of the LLM/generative AI, the architecture of IT could deliver continuous refinement. This adaptability ensures that it remains relevant, accommodating emerging manufacturing methodologies and integrating the freshest insights from physics and chemistry. Also, using GitHub, one could get support/assistance from other developers to improve the accuracy of the code. PROJECT DURATION and COST: Proposals exceeding these limits will not be evaluated. PHASE I: Not to exceed a duration of 6 months and cost of $100,000. PHASE II: Not to exceed a duration of 18 months and cost of $1,000,000. PHASE I: Phase I will consist of delivering a TRL level 3 proof of concept that will include the design of an IT system that calculates value-add, non-value-add, and the total energy for a minimum of 15 discrete manufacturing processes. In addition to calculating the energy required for various manufacturing processes, it should include the greenhouse gas emissions from the energy use during value-add and non-value add processes. This system should plan to meet all the DoD physical and cybersecurity requirements. Collaboration with a relavant DoD Component organzation/supplier (e.g., DoD lab and/or prime contractor) and one or more relavant DoD weapon system supply chain participants or other suitable organization is reccomended. PHASE II: Phase II will build a working prototype system based on the Phase I design. The prototype should address the identified discrete manufacturing processes from Phase I. Furthermore, it should be used to confirm the estimates and provide preliminary cost and pricing data. A business case will be generated using both DoD and commercial markets. Collaboration with a relavant DoD Component organzation/supplier (e.g., DoD lab and/or prime contractor) and one or more relavant DoD weapon system supply chain participants or other suitable organization is required. PHASE III DUAL USE APPLICATIONS: At this point, no specific funding is associated with Phase III. Relationships developed and progress made in Phase I and Phase II projects should result in the ability to produce to DoD orders and organic growth of business from there. REFERENCES: https://publications.anl.gov/anlpubs/2010/10/68288.pdf https://us-west-1-02900067-inspect.menlosecurity.com/safeview-fileserv/tc_download/beb9a1d4aeb1d8ce2d2fdfb4fb637f52e8537f85f3ac8225281c2e67de92cdbc/?&cid=NC949DF232947_&rid=53a12180f385ba1d5ce1e51c044cdc97&cl=9LEI8II9B3b&file_url=http%3A%2F%2Fweb.mit.edu%2Febm%2Fwww%2FPublications%2F9_Paper.pdf&type=original KEYWORDS: discrete manufacturing, energy use, value and non-value add, IT system, Large Language Model (LLM), Generative AI