Waters Corp.

Waters Inc. was conducting research on new pressure transducer designs and I was given the task of figuring out optimal dimensions for the thickness between the inner tubing and gauge surface, as well as finding optimal strain gauge placement locations. This distance between the surface and inner tube would allow a specified maximum strain value of no more than 1800με on the gauge surface given an applied pressure specification of 25,000 psi through the transducer. I iterated the design's thickness down and ran each in ANSYS software until the surfaces maximum strain value fell slightly below the allowable 1800με. Plotting the strain from the the center of the gauge surface to the outer edge, I was able to find the optimal strain gauge locations by finding the distance from the center of the gauge surface where the strain dipped the lowest below zero. Knowing the specific strain value of this location the voltage signal output could be calculated and compared to the original design to confirm that this new design was more optimal for use.

Experimentation had previously shown that a bending stress on a transducer's inner tubing can greatly affect its signal output depending on it's pressure gauge layout. This experimentation involved putting a slight bending force on the inner tubing and seeing if there was a change in signal output; however, no formal test or test plan had been created to quantify these issues. My task was to design a test that could be consistently used for testing future gauge layouts against bending stresses in multiple directions. To test these transducers, I designed and 3D printed a fixture that could be turned on any side when the transducer was mounted. I attached long socket head screws that would fit the threads on the tranducer's and attached weights to the end of both sides. I then processed the transducer's output signal through LabVIEW and graphed it with a Python script to compare how the bending in each direction interfered with the output signal. I wrote a test plan about evaluating future transducers and a report to document the transducers I did test. I created another Python script to analyze basic operational tests of these transducers. This script loops through any number of LabVIEW data files in a folder and reports the maximum accuracy, linearity, repeatability, and hysteresis errors for each. All of these values are recorded in a master Excel file with a pass/fail characteristic, so that different transducer designs can be compared.

I performed Instron compression testing for a variety of applications including material research and ferrule un-crimping forces. The material research was conducted on a type of plastic that was failing prematurely in a chromatography device. The material came from different batches of injection molded plastic. I performed compression tests with many samples from each batch to find out whether there was a statistical significance in their difference of yield loads and elastic moduli.

One project I conducted included slicing epoxy-filled pump heads with a waterjet to study their cross-sections. The first task I completed was the design of a fixture in SOLIDWORKS that would hold the sliced pump heads with their cross-sections parallel to a KEYENCE Digital Microscope that would captor images of the sections. I 3D printed a fixture to house multiple epoxy shapes, and took pictures of all available pump head surfaces with the microscope. I was also given the task to account for possible issues in the waterjetting process. The images confirmed that the waterjet did not perfectly cut the pump heads down the middle. The cross-sections could be offset or at a slight angle from the center axis of the pump head. To combat this, I wrote a Python script with the module ginput that allows the user to directly make changes to the cross-sectional image by following steps in the script to correctly dimension certain parts of the pumphead. The script then reforms the image, expanding or contracting certain parts to be scaled correctly in comparison to the other pump head parts in the image. The script requires 10 steps of data inputs including dimensioning multiple parts of the pump head and calculating the average distance between clicks to find the center axis. The result of the script is the same cross-sectional image that has been slightly contorted and rotated to match the dimensions had it been sliced perfectly down its center axis.

Because so much of the work being done on devices involves liquid flow, much of the testing comprised of checking for leaks. One of the tests I performed consisted of qualifying a back-locking ferrule design. These ferrules were tested at many pressures up to 19,500 psi. The ferrules met the required performance against leakage by not exceeding the maximum allowable leakage rate of 0.02 μL/min for all trials. Finally, I created a test report on these ferrules to be used as a reference for future comparisons. Another test incorporated validating a new gasket design as a fix for past leakages in which a solvent was getting through to the electronics in a device. For this, I placed thermal paper around the electronics in the device that would turn blue if it came into contact with the solvent. This testing was done while the device was mounted at multiple angles to determine if angle affected leakages. I wrote a final report displaying all of the different tests and the findings for each to be reviewed and determine whether the updated gasket design would be put into use.