Dataset of Agar Biopolymer Films: mechanical properties

We study the tensile properties of biopolymer films made from the seaweed biopolymer agar. We test whether a process with low capital costs can result in materials that compare favorably to commercial plastics. Tensile properties are key to assessing the applicability of biopolymer films to a range of single-use packaging, such as lightweight plastic bags and plastic films. We identify formulations that compare favorably to commercially available plastics, demonstrating the potential to address standard product niches. The complete set of formulations and their associated properties can be used to understand the broad range of performance characteristics that can be obtained using one process. Formulations were developed using a Design of Experiments (DoE) approach, following a full factorial mixture design using the methodology 2k where k is the number of factors considered at 2 levels. A neural network was trained to model the relationship between the mechanical properties of test samples and their composition. Our aim was to demonstrate the importance of such methods for modeling the properties of bioplastics produced at local and regional scales where data may vary with heterogeneous sources of feedstock and local process conditions. The process of fabrication consisted of adding agar to a mixture of glycerin and water and heating to a temperature of 90°C using a magnetic stirrer and heater. The mixture was then poured into silicone molds, where it dried over a period of approximately three days. 32 formulations were fabricated and tested, with varying concentrations of the ingredients agar, glycerin, and water. Tensile tests were performed on an INSTRON universal test machine. Mechanical properties analyzed were ultimate tensile strength, Young's modulus, and elongation at break. A uniaxial tensile test was used to evaluate the stress and the deformability of the material. An average of 5 specimens were tested for each formulation, and the resulting data was normalized and interpolated with deformation averaged to obtain the main curve. The dataset contains the following: (1) the principle table with the formulation mixtures, calculated tensile properties, dimensions of each sample prepared for testing, and relevant process variables; (2) a diagram specifying the dimensions for the die used to cut each sample for tensile testing; (3) the abbreviated table used as input for the neural network including the sample formulations and tensile properties, and (4) the python script used for running our neural network model