The Bowles Fluidics’ automotive washer nozzles are being prepared for the cross-sectional scanning process. Source: Bowles Fluidics

Bowles Fluidics (Columbia, MD) applies the science of fluid logic circuits to put liquids in motion for a variety of consumer and automotive products. The advantage of fluidics is that a spa jet or windshield washer nozzle will oscillate a stream of liquid without the use of moving parts. The challenge is that a fluid circuit demands precision in its design and production.

Circuits are designed with tiny, complicated geometries that must be injection molded to tight tolerances. Small deviations cause short circuits that degrade performance, create leaks or prevent oscillation. These circuit problems are difficult to diagnose and resolve since they lie within the molded, assembled parts. “With conventional inspection tools, we always seemed to be chasing the problem. We were never quite sure precisely what were the dimensions of the internal geometries, due to the difficultly in quantifying internal measurements,” says Alan Romack, senior project engineer for Bowles Fluidics’ automotive products.

Chasing the problem meant sneaking up on the repair. After measuring sample parts from the molding run, the mold would be modified by removing only a small amount of steel because of the limited inspection information. This conservative approach avoids taking away too much metal and scrapping a precise, expensive component of the tool. Samples would then be molded and examined, and the process would repeat until the problem area was resolved. This iterative process was difficult on product managers since there usually was not extra time built into program schedules.

The cross section shows a sink caused by the non-uniform profile around the spherical pocket of an adjustable-aim nozzle. Source: Bowles Fluidics

Complex Nozzles

To get ahead of this kind of problem and hone in on the fix, the company needed more information and better insight into what was happening on the inside of its molded parts. For this, it invested in a CSS-300 Cross-Sectional Scanning system from CGI (Eden Prairie, MN). Soon after its installation, the CSS system was put to the test for the company’s product lines, including automotive nozzles.

Adjustable-aim nozzles are more complex than traditional, fixed-aim designs. To gain adjustability, the nozzle incorporates a spherical seat that makes injection molding and tool making even more difficult.

The non-uniform profile around the spherical pocket makes the wall thickness inconsistent, which then makes “sinks” a bit unpredictable. Since a gap as a small as 0.001 inch will cause leaks and streamers, a little sink can be a big obstacle, as it was on a recent headlamp washer project.

This new, adjustable-aim nozzle had a perfect design, yet the first molded samples did not perform well. “When we tested the molded parts, we were really disappointed. There were big streamers,” says Romack. “Without detailed information from the samples, this is the kind of project that could have turned into a big problem. So, it was once again time to put the CSS-300 to the test.”

The CSS-generated 3-D point cloud documents internal and external dimensions of the headlamp washer nozzle. Source: Bowles Fluidics

The cross-sectional scanning technology provides first-article inspection of internal and external features from high-density point clouds. To get these measurements, the first step was to encase the nozzle in a slow-curing plastic resin. Then the encased, or “potted,” part was placed in the CSS-300 inspection system, which is fitted with a precision fly-cutter tool that machines ultra-thin (0.001 or 0.002 inch) layers. As each layer was milled off, an optical scanning system captured the newly exposed 2-D profile at a resolution of approximately one million pixels per square inch. The milling and imaging process repeated until the part was consumed. The collection of 2-D images was then processed to build a 3-D point cloud that detailed the exterior profile and internal channels.

Romack imported the point cloud and the original computer-aided design (CAD) model into InnovMetric’s (Quebec City, Canada) PolyWorks software application. Using PolyWorks Inspector’s part-to-CAD comparison tool, he created a 3-D color map that visually depicted the variance of the molded part from its design. This allowed them to see exactly where the sinks were located and how big they were. The color map showed bright blue at the site of the sink that immediately drew attention to the problem area. “Looking at the color map, we knew exactly what to do,” says Romack.

A 3-D color map shows the exact location of the sink that caused fluid streamers. Source: Bowles Fluidics

Using poly lines created from cross sections of the point cloud, the tool designer adjusted the CAD model to fix the problem in the precise area of the sink. Having a complete inspection picture, he was able to ignore deviations that were not critical to function. The pins for the two-cavity mold were then sent off for machining, the tool was assembled and a second batch of sample parts was molded.

“We assembled the samples nozzles, and they worked flawlessly,” says Romack. “With only one revision cycle, we were able to mold parts that worked perfectly. We made the tooling change, machined the pins and we were done. We did not have to sneak up on the fix for this one.”

Following the successful test, the nozzle was rescanned with the CSS-300 to document the part and complete first article inspection. Romack says, “We did a side-by-side comparison of the color maps from the first and second sample runs. It was astonishing.” Where the original sink was located, the seat was perfect and surrounding areas were unchanged, as the tool designer had planned. “The color map went from blue [sink], to green where it counted and back to blue,” he says.

A cross section shows that the nozzle’s sink was eliminated in one revision cycle. Source: Bowles Fluidics

Finishing the Job

Bowles Fluidics was impressed with the one-shot solution that the CGI system allowed. It took just nine working days from the inspection of the first samples to the testing of the second batch of parts to dial in a working tool for the headlamp nozzle. “According to the director of engineering, this problem-solving exercise could have taken close to two months to get right, if we didn’t have the CSS-300.

“We have validated our CSS-300 purchase with this one job,” says Romack.

On the nozzle project, and the others like it, Bowles Fluidics has discovered some unexpected benefits from the CGI system. With the high-density point cloud, the engineering team can measure any feature of interest at any time.

“With the point cloud, we zoom in on a feature and measure it, giving us any dimension we want,” Romack says. “We provide the measurements we knew were needed and anything that comes to mind as we are investigating the parts.” With the archived point cloud, Bowles also has an exact definition of its parts as they were originally manufactured. This allows them to troubleshoot any future problems by comparing nozzles to the real-world parts from the newly launched tool, not to the theoretical model in the CAD file.

The point cloud also is used to investigate variances between mold cavities on a multi-cavity tool. “With the CSS-300, we can document the results from each cavity. We no longer assume that the results are the same,” says Romack. Instead of averaging across multiple cavities, each is independently and precisely tuned to mold parts to the tight tolerances needed to make a fluid logic circuit work.

“The primary driver for the purchase of the CGI CSS-300 was to help us diagnose and correct molding issues,” says Romack. As shown by the headlamp washer project, the system has done that by taking the inaccuracies in measurement out of the process, which accelerates the product launch. He concludes, “You don’t have to be concerned about the quality of the dimensions anymore. Now you can focus on the solution.”

CGI Inspection
(952) 937-2005


  • An optical scanning system captured the newly exposed 2-D profile at a resolution of one million pixels per square inch.
  • It took nine working days from the inspection of the first samples to the testing of the second batch of parts to dial in a working tool for the headlamp nozzle.
  • The cross-sectional scanning technology provides first-article inspection of internal and external features from high-density point clouds.