Spatial strain measurements using a strain-sensing grid patterned from nanocomposite films

In recent years, interest has grown in the development of sensing skins for structural health monitoring (SHM) with electrical impedance tomography (EIT) used to image skin properties. The computational e ort associated with the inverse solver of EIT is very large and sometimes the final reconstructed strain map derived does not correspond to the true state of the system. To reduce the large computational effort associated with EIT reconstruction of large unpatterned thin films, this study fabricates a patterned spatial strain sensor made from single-walled carbon nanotube (SWCNT) nanocomposite films. The thin nanocomposite strain sensing films are fabricated on a flexible polyimide substrate by using a layer-by-layer (LbL) deposition process. Rather than using a plain/unpatterned film as previously proposed in EIT-based approaches, a grid of piezoresistive nanocomposite strip elements are patterned. The 2D grid arrangement of the nanocomposite sensing elements is achieved using optical lithography. Metal electrodes are deployed at the boundary nodes by physical vapor deposition (PVD) and are connected to wires for controlled current injection and electric potential measurements. In order to infer the strain distribution of a structure using the patterned strain sensing skin, there is a need to have a robust inverse solver. Ohm's and Kircho's laws are used to derived the Neumann-to-Dirichlet map of the rectangular resistor network. The Neumann-to-Dirichlet operator is represented by a matrix which is a function of the resistance distribution. The inverse solution obtains the resistance of each grid element by updating the Neumann-to-Dirichlet operator to drive convergence between the boundary potential measurement taken of the film and those predicted by the forward solution. The inverse solver has been validated by simulations and experiments of a square resistor network of 3 by 3 (node by node) and 4 by 4 resistor grids. Preliminary results show that the proposed inverse solver is accurate for 2D strain measurement using the patterned strain sensing grid.