NIO launches 100 kWh battery with flexible battery upgrade plans; cell-to-pack technology
PNNL’s hydrothermal liquefaction shows promise for turning toxic algal blooms into biocrude

ANL team generates 3D images of fluid flow inside steel fuel injector nozzle to improve design models

Scientists at Argonne National Laboratory have used the Advanced Photon Source (APS), a DOE Office of Science User Facility located at Argonne, to generate the first 3D images of the fluid flow inside a steel fuel injector nozzle.

The high-speed 3D visualization will help engine manufacturers and suppliers improve the design models that are used to create fuel injectors. The requirement for higher fuel efficiency and lower emissions has emphasized the need for tight design and manufacturing processes with higher injection pressures and smaller orifices. As an outcome of the experiments at the APS, the Argonne engineering team also created open-source software to help others analyze images acquired in low-light conditions.

The experiment, detailed in a recent open-access paper in Scientific Reports, was developed by Christopher Powell, principal engine research scientist at Argonne. Powell, postdoctoral researcher Aniket Tekawade and others used X-ray imaging to discover that manufacturing imperfections resulted in low-pressure regions inside the nozzle as fuel was sprayed. This caused the fuel to vaporize in these regions, forming pockets of gas inside the nozzle that can ultimately erode steel.


A surface representation derived from high-resolution X-ray computed tomography of the fuel injector (ECN Spray C37) used in the study shows an asymmetrically drilled hole and a 20 μm wide ridge on the nozzle wall at the inlet. Slice from synchrotron X-ray CT scan (left), full iso-surface (middle), zoom-in view of inlet to nozzle (right). Tekawade et al.

The morphology of the internal flow reveals strong flow separation and vapor-filled cavities (cavitation), the degree of which correlates with the nozzle’s asymmetric inlet corner profile. Micron-scale surface features, which are artifacts of manufacturing, are shown to influence the morphology of the resulting liquid-gas interface. The data obtained at 0.1 ms time resolution exposes transient flow features and the flow development timescales are shown to be correlated with in-situ imaging of the fuel injector’s hydraulically-actuated valve (needle).

—Tekawade et al.


Slices from the fully reconstructed CT volume (left) show flow separation and cavitation leading up to the nozzle exit (nozzle length ~ 1  m). As per the coordinate system, x = 0 μm is at the nozzle tip. The inset figure shows histogram of voxel intensity in the slices, indicating a distinct binarizable intensity map, with high and low intensity corresponding to liquid and gas phases respectively. 3D volume rendering of the liquid phase is shown on the right (blue-green), further illustrating the wrinkled nature of the liquid-gas interface. Tekawade et al.

The team automated many facets of the experiment; from taking X-ray images to rotating the fuel injector nozzle so that images could be taken at multiple angles. Using production fuel injectors allowed them to perform experiments at the very high pressures typical of diesel engines, but led to the challenge of seeing through the steel body.

This was different from past fluid flow experiments by others, which typically use transparent plastic or glass nozzle replicas at low pressures to watch the fluid flow inside. Actual fuel injector pressures would break glass or plastic replicas.

Even with one of the most powerful X-ray sources on the globe, almost 99% of all the X-ray photons from the source are absorbed by the steel body and we are left with only 1.5% of the photons to show the fluid flow inside the injector. Moreover, the leftover photons possess energies so high that the liquid fuel is 99.97% transparent to them.

—Aniket Tekawade

The challenge of scale included processing the vast quantity of X-ray images taken over the 30-hour experiment (more than 100,000 images, equivalent to 2 terabytes of data) to improve poor contrast caused by the small proportion of X-ray light that could pass through the steel.

Image processing tools were created and used by Tekawade to process the X-ray images into 3D maps of fluid flow. To help others use images acquired in low-light conditions, Tekawade created open-source software, developed from deep learning algorithms.

Tekawade’s open-source software, CTSegNet, can be accessed at the following link:

Funding for this research was provided by the Department of Energy’s Office of Energy Efficiency and Renewable Energy, Vehicle Technologies Office.


  • Tekawade, A., Sforzo, B.A., Matusik, K.E. et al. (2020) “Time-resolved 3D imaging of two-phase fluid flow inside a steel fuel injector using synchrotron X-ray tomography.” Sci Rep 10, 8674 doi: 10.1038/s41598-020-65701-x


The comments to this entry are closed.