Machine Learning Database to Guide Development of Low Flammability Polymer Matrix Composites

Navy SBIR 23.2 - Topic N232-106
ONR - Office of Naval Research
Pre-release 4/19/23   Opens to accept proposals 5/17/23   Closes 6/14/23 12:00pm ET    [ View Q&A ]

N232-106 TITLE: Machine Learning Database to Guide Development of Low Flammability Polymer Matrix Composites

OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Sustainment;Trusted AI and Autonomy

OBJECTIVE: Develop an active machine learning (ML) database to aid the Navy in the development of polymer matrix resins and composites that have low flammability. as demonstrated under ASTM E1354 (heat release rates) by cone calorimeter. The Navy has very strict flammability requirements for composite materials to qualify for use below deck (MIL-STD-2031), which must meet metrics for time to ignition, maximum heat release rate, and smoke density (IAW ASTM E662).

DESCRIPTION: Use of polymers and composites below deck on a ship is very limited because the polymer matrix resins potentially provide fuel to a fire. Use of composites in general could save weight and reduce maintenance. In applications such as pressure vessels, there is potential to save costs as well. However, the epoxy matrix resins typically used are too flammable and the composite vessels will not meet Navy flammability requirements. Polymer resins that have reduced flammability typically leave more char when burned. They are highly crosslinked materials that are brittle and must be cured at higher temperatures making them more expensive than metal pressure vessels. Addition of flame retardants to the epoxy resins can reduce their properties.

A composite is a system composed of a matrix resin, reinforcement, and possibly other additives. The reinforcements and additives can improve the flammability performance of the composite by restricting oxygen flow to the resin as an inert filler or as an active filler promoting the formation of a blocking layer. The mechanical properties of a polymer composite (i.e., modulus, strength) can be predicted based on resin properties, fiber/filler properties, and fiber volume fraction and orientation. Addition of flame retardants provides a new variable as generally these decrease mechanical properties, though some types could enhance properties.

Working through these variables to identify composites systems that could be used below the deck on Navy ships has proven to be difficult. A ML database could help and could make use of the fairly plentiful data on composites as building materials to predict avenues for the Navy to pursue.

ML databases can be constructed such that they can take many inputs, either experimental or computational, which may be used directly as descriptors to correlate to a desired predicted property, or used to calculate a descriptor through physical or empirical relationships. It is a learning process to see which descriptors yield or correlate to predicted properties which best match experimentally determined properties. When this happens, then reverse design is possible. With this learning process in mind, we would like to start at a fairly simple level with composite component materials on the input side and Navy performance metrics on the output side to evolve an effective ML database for composite materials with low flammability that meet Navy performance needs (modulus, strength, thermal stability). Work will start in Phase I with trying to estimate the flammability of a composite. The Navy has performance requirements based on ASTM E1354 testing with limits given in MIL-STD-2031 [Refs 1-2].

PHASE I: Develop an expandable ML platform that can use: (1) literature data and; (2) first principle calculations to predict the flammability index from the chemical structure of a neat resin. Develop an approach toward predicting ASTM E1354 Cone calorimetry results for maximum heat release rate, time to ignition, and smoke density.

PHASE II: In year one of the Phase II, composite properties will be added based on typical glass fiber and carbon fiber compositions/geometries/volume loading of Navy composites and commercial structural composites. In consultation with the Navy, neat resin and composite samples will be tested to ASTM E1354 and the data will be used to both evaluate the ML database and to add to it. In year two of the Phase II, common flame retardants will be added to neat resins and composites in a second round of ASTM E1354 testing, again to test this capability of the ML database and to add to it. In Phase II Option, if exercised, mechanical properties of the composites with resin/fiber/flame retardants could be added or other ML database maturation based on discussions with the Navy team.

PHASE III DUAL USE APPLICATIONS: Make the system user friendly, allowing the users to add their own databases and to prioritize various data sources already incorporated into the model. Transition the platform to the technical warrant holder for flammable structural materials and to material engineers trying to improve materials.

The database is dual use as low flammability structural materials are needed for commercial and residential buildings, for aircraft and automobile interiors, and other applications in addition to being used on pressure vessels, storage tanks, hatch doors, and so forth below deck on Navy ships.


  1. "MIL-STD-2031 Fire and Toxicity Test Methods and Qualification Procedure for Composite Materials Systems Used in Hull, Machinery, and Structural Applications inside Naval Submarines."
  2. "ASTM E1354-22b Standard Test Method for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter."
  3. ASTM E662-21ae1 Standard Test Method for Specific Optical Density of Smoke Generated by Solid Materials."

KEYWORDS: ASTM E1354; composite; heat release rate; machine learning; database; flammability


The Navy Topic above is an "unofficial" copy from the Navy Topics in the DoD 23.2 SBIR BAA. Please see the official DoD Topic website at for any updates.

The DoD issued its Navy 23.2 SBIR Topics pre-release on April 19, 2023 which opens to receive proposals on May 17, 2023, and closes June 14, 2023 (12:00pm ET).

Direct Contact with Topic Authors: During the pre-release period (April 19, 2023 through May 16, 2023) proposing firms have an opportunity to directly contact the Technical Point of Contact (TPOC) to ask technical questions about the specific BAA topic. Once DoD begins accepting proposals on May 17, 2023 no further direct contact between proposers and topic authors is allowed unless the Topic Author is responding to a question submitted during the Pre-release period.

SITIS Q&A System: After the pre-release period, until May 31, (at 12:00 PM ET), proposers may submit written questions through SITIS (SBIR/STTR Interactive Topic Information System) at by logging in and following instructions. In SITIS, the questioner and respondent remain anonymous but all questions and answers are posted for general viewing.

Topics Search Engine: Visit the DoD Topic Search Tool at to find topics by keyword across all DoD Components participating in this BAA.

Help: If you have general questions about the DoD SBIR program, please contact the DoD SBIR Help Desk via email at [email protected]

Topic Q & A

5/25/23  Q. 1. Is it desired that first principal calculation simulations and frameworks are directly accessible in the Phase 1 POC, or can existing calculation result data be used for Phase 1?
2. Do the TPOCs have a first principle simulation framework in mind? Generally, different frameworks excel at different calculations, and this may inform our approach.
3. Do the TPOCs have specific parameters from calculations that they would want in the database (e.g. free energy, heat of formation, etc)?
4. Do they want the system to sift through raw outputs of the first principles simulations to obtain certain parameters?
5. Do you have an existing Program of Record(POR) or any future potential POR?
   A. 1. In phase I a ML platform should be developed and populated with publicly available data and a computation framework to generate such data should be developed. The result in phase I should be a ML platform that can predict ASTM E1354 results based on neat resin structure.
2. There is no preferred simulation framework
3. In terms of predicting ASTM E1354 results for maximum heat release rate and time to ignition this could be done at an irradiances of 75 and 100 kW/m2. Computing more fundamental parameters will be necessary to predict the ASTM 1354 results
4. See answer to number 3
5. There is a transition pathway but at this time not a future POR
5/16/23  Q. Can you suggest any datasets or databases that contain compounds/polymers along with their flammability index, heat release rate, time to ignition, or smoke density?
   A. NIST Fire Institute or the FAA
5/11/23  Q. 1. By "ML database", do you mean a database with built-in machine learning algorithms?
2. In Phase II, will the NAVY provide the data for testing to ASTME1354 or do we need to generate the data ourselves ?
   A. 1. Yes.
2. The Navy will not be providing data to populate the database.

[ Return ]