Research Accomplishments of Gregory A. Campbell, PhD

During my 38 year career at General Motors, Mobile Chemical, Clarkson University, and now as a consultant, I have performed research on problems that were industrially important. In many of these problems, new fundamental understanding of the process or problem had to be developed. The next sections summarize the most important research accomplishments of my career.

In the area of polyurethanes several fundamental research accomplishments were developed that positively impacted the business at General Motors (GM). First, a mechanism and quantitative simulation was developed for the development of skin in the integral skin polyurethane (PU) foam process. We determined that the skin was the result of local condensation of the cell gas due to the interaction of the local pressure and temperature as predicted by the Clausius-Clapeyron equation. This thermodynamic mechanism was driven by the heat transfer dynamics causing local cooling near the cool mold wall. A dynamic simulation was produced and the results predicted the skin formation accurately in a number of molds. The results were presented over two years at the Gordon Research Conference on foam and published so that the industry could use them. The developed mechanism was later used by the polyurethane chemical manufacturers to optimize foam formulations to produce integral skin foams and to specify the pressure and mold temperatures to get consistent skins on complex parts.

The second area of fundamental advance in PU foam was the development of a technical solution to the problem of humid age compression set. The problem occurred when GM decided to introduce a “cold cure foam” for automotive seats. An important characteristic of automotive seat foam is it resistance to taking a permanent reduction in thickness while in use, this is called compression set. The specification was 15% compression set which was easily met by the current “hot cure” foam, but could not be reduced below 70% by the three primary foam chemical suppliers during a 3 year concentrated effort using a “cold cure” method. I directed a group of researchers and we addressed this issue. We determined that the high humid age compression set was the result of a secondary reaction of a toluene diisocyanate (TDI) crystalline urea that was formed during the early part of the foaming process. The TDI crystalline urea was formed due to the amine catalyst used in the foam formulation. The crystalline di-urea subsequently hydrolyzed and dissolved in the polyurethane matrix and formed new crosslinks thus causing the set. We then conducted fundamental kinetic studies using different catalysts and determined that a small amount of tin catalyst promoted the urethane reaction early in the foaming process so that when the urea formed while producing the carbon dioxide, used to reduce the foam density, it was bound in the matrix and thus was not available to form unwanted crosslinks as the foam was used. This research allowed GM to produce foams the met use specifications with less than half of the energy cost of a “hot foam.” The chemistry and process was used by other industries to produce high volume molded foam parts while reducing the energy input and cost.

A fundamental question at the time was what causes the “ride” in deep seat foam seats to degrade over time. A fundamental investigation was conducted and the cause of the loss of resilience in the foam was found to be compressive creep. An experiment using the maximum size sample that would fit into an electron microscope was carried out under relatively low magnification. It demonstrated that as the foam compressive strain is increased the struts buckle, and the increase in local strain of the foam matrix causes the creep to accelerate to the point that in a relatively short time the struts are interfering with the motion of other struts, leading to the “hard” ride as the foam ages. The research led to foam formulations that had cell sizes that produced the struts that would buckle at the desired strain and with enough inter strut membranes to minimize strut to strut interference during most use conditions.

Also at GM, an electro-deposition process was used to put primer on metal car bodies. Periodically, the corrosion resistance of a group of cars was extremely poor and caused severe customer complaints and warranty costs. I conducted a fundamental investigation by attaching micro-thermocouples to the panels. We determined that the root cause was local boiling of the paint at the surface of the steel panel. The cause of the local surface boiling was excessive current flow at the end of the paint deposition process. Fundamental process simulations indicated that the problem was the high conductivity of the paint film. Further investigation using paint from a tank that had just been filled with new paint showed that the current in the deposition process decreased rapidly as the paint was deposited and there was no subsequent corrosion problem with panels coated in this paint. Further investigation indicated that the “bad” paint was contaminated with ionic material carried into the tank from the phosphate process for treating the steel before it was painted. A system was developed to monitor the ionic level in the tanks and remove them when they reached a level that adversely affected the corrosion resistance of the painted auto body. Implementing the technology resulted in a dramatic decrease in corrosion in the cars and a reduction in customer complaints.

One of the most interesting projects that I worked on was the development of the neck in anthropomorphic dummies used in auto crash tests. The research was in collaboration with the Bioengineering group at GM. Accurate data from anthropomorphic test dummies that correlates with real world human data is absolutely necessary when developing auto safety paradigms. At the time, the neck in the available dummies did not represent the effects of the ligaments and muscle in the human neck as they responded to the head acceleration during a crash. GM had excellent data on these forces and accelerations that are developed in the human neck during a crash. How these data were acquired is another story. One of the bioengineers had developed a neck simulator which is now in production as many of us have seen it on TV when auto safety is discussed. The neck simulator is a series of discs, think vertebrae that work in compression instead of the tensile forces that are actually developed by the muscles and ligaments due to the human head motion during the crash dynamics. I developed a polymeric material and a specific shaped disc that fit between the “titanium vertebrae.” The molded shape and the modulus of original polyurethane provided excellent correlation for head acceleration and anthropomorphic data when the dummies were put through crash simulations on the test sled. Once we knew the shape and modulus needed we developed a compression molding technique to make the parts. The design and the manufacturing details were given to the US government and this work is the basis of our current crash test dummies.

In the mid 70’s I added high-speed analog control to the injection molding machines at GM, and when I moved to Mobil Chemical I continued using this kind of control. Analog control occurs in real time and thus there is no “digital delay” in the control system. We added high-speed process monitoring on the machines, collecting 31 channels of data every 8 milliseconds. We used infrared temperature monitoring in the barrel and was able to see the temperature transient across the plasticator screw flights and able to use the flight crossing the probe as a timer for screw rotation. We developed a set of software that monitored the temperature and pressure variables in the nozzle and mold cavity such that the number of parts needed to get good physical properties was reduced from 20 to 5. The key to this reduction was to have the data in real time on the computer screen and we did not take parts until the lines of the variables overlapped. A number of papers relating to injection molding screw design, temperature dynamics and part properties were published particularly relating to part stability and recoverable strain in the molded parts as a function of process parameters and the amount of regrind in the hopper. It is important in molding large or high tolerance parts to minimize dynamic changes in dimensions at use temperatures. These advances in understanding of the interaction of screw design, melt temperature, injection rate, injection pressures, and part properties improved molded part reliability throughout the industry and reduced the cost of evaluation of new resins.

I developed with T. A. Huang in the mid 80’s a method for determining the dynamics of stretching for the blown film process using a video camera. The paper that describes this research is one of the fourth most cited paper from the journal, Plastic Film & Sheeting, that it is published in. The technique is still used today by researchers that are investigating the blown film process. After joining Clarkson University I carried out about 20 years of fundamental research on blown film. Our first discovery was that the model for blown film, which can be considered based on generalized spherical physics, always closes the film into essentially a spherical shape when simulated above the frost/freeze line. The film produced in production and laboratory equipment is essentially straight above the freeze line. This is apparently why the literature always terminated the simulation at the frost/freeze line. We published the first model that addresses this issue by incorporation of a yield stress in the rheological equation. When modeling blown film the temperature in the film as it cools is affected by the crystallization and the internal and external air blowing (cooling) on the film surfaces. We investigated the dynamics of the air from the air ring regarding turbulence and the thickness of the boundary layer. This led to a theoretical model that predicted the dynamic pressure on the outside of the bubble could be of the same order as the internal bubble pressure. When these two pressures are equal the bubble tends to become unstable. Internal bubble cooling (IBC) increase the pressure inside the bubble and thus higher velocities can be used to cool the outside of the bubble. These higher velocities lead to higher external dynamic pressures. We developed a technique using multiple temperature black body sources to be able to get the average bulk temperature and we were one of the first laboratories to recognize that 3.43 µm infrared probes would give us the temperature in the outer 25 µm of the film. Thus we had two temperature measurements that we could use to verify film blowing models. This paper is the third most cited paper published in Plastic Film & Sheeting. This also led to the development of a heat transfer analysis and model that takes into account the change in boundary layer thickness. This boundary layer thickness increases due to the distance the air flows up the bubble and we discovered decreases due to the boundary layer being stretched as the bubble increase in diameter. With this model we were able to show that the measured temperature in the blown film was always higher above the freeze line than the models predicted. We hypothesized that this may be due to crystallization continuing above the temperature plateau. This led to development of an on line small angle light scattering system to follow the polyethylene (PE) crystallization dynamics. To our surprise not only did the crystallization continue above the point of the bubble where the temperature plateau was found but the majority of the crystallization occurred above the freeze line, which is at about the same position as the temperature plateau. All reports in the literature prior to this assumed that the majority of the crystallization occurred in the plateau which is just below below the freeze line. When this physics was added to the models, a very good agreement was found between the infrared temperature measurement and the models. Now that the temperature and crystallization dynamics were understood we determined that machine direction (MD) and transverse direction (TD) properties could be presented and correlated on a single surface by using reduced strain and reduced strain rate as the independent axes and the property as the dependent axis. This leads to the potential of taking data on a lab line and prediction of the properties on a production line if one can predict the stretching dynamics, temperature, and crystallization dynamic. All of these advances are published in the open literature. Much of this work is summarized in the two books I co-edited with Toshitaka Kanai; “Film Processing” published in 1999, and “Film Processing Advances” published in 2014. The video analysis technique is still used extensively by those that need to evaluate the stretching dynamics of the blown film process. The advances in understanding of the bubble external pressure, temperature dynamics, and particularly the new understanding of the PE crystallization dynamics have aided industrial R&D and Technical Service professional to reduce the time to develop new resins and to optimize the development of physical properties by controlling the process variables such that the optimum frozen in strain is developed as the film is produced.

In the late 1980’s my Clarkson Research group discovered a flaw in the literature model for single-screw extruders. The original model development occurred in 1922 by Rowell and Finlayson when they transformed the metering channel of the screw into model with a stationary screw and a moving barrel reference. At the completion of the analysis, Rowell and Finlayson failed to move the reference frame back to the laboratory reference frame of a moving screw and stationary barrel. This flaw was not discovered until the earlier 1990’s by my group. When the original model was move back to the laboratory reference frame, it predicted flow in the wrong direction; i.e., flow was towards the feed end of the machine. The flaw created numerous issues with how single-screw extruders fundamentally work. These include the mechanism for flow in the metering channel, energy dissipation and thus material temperature, melting, and solid bed breakup. For example, the original model had the motion of the barrel dragging the material forward in the metering section of the screw as the mechanism for the main part of the flow. This flow was known as drag flow for the years that followed. My research has shown that the mechanism for this flow is the motion of the forward motion of the flights and the backward flow due to the motion of the screw core. I have worked to develop a more functional understanding of the dynamics and physics of the single-screw extruder and more complete models for single-screw extruders. All of the experimental data and theory have been published in the open literature. My approach is to take data that addresses the open questions in the literature and then develop models that allow us to correlate the data in a manner that the models can be used to predict well beyond the confines of the data. The result of this work is an understanding now that an alternative analysis using different boundary conditions leads to a model that addresses in a manner that is consistent with data the deep channel problem, the fluid directional flow of the model when transferred back to the laboratory frame which is of course where we take the data, the temperature rise is now consistent with data taken again in the laboratory frame, a new melting model consistent with data analysis, and a new solid conveying model that much better represents what we observe in the lab experiments.

In order to design a screw properly, the designer must be able to predict the rate, extrudate pressure, and extrudate temperature. This was a very difficult task using the existing models based on barrel rotation boundary conditions. For example, for a polycarbonate (PC) resin extrusion process barrel rotation methods predicted a discharge temperature that was about 8oC higher than that for screw rotation methods. The difference of 8oc does not appear to be that great, but it can translate to significant differences in expected rate for a commercial application. The original melting model developed by Bruce Maddock and numerically quantified by Tadmor and Klein utilizes barrel rotation boundary conditions and a rearranging solid bed. The model appeared to work well for the first 25% of the melting process, but then deviated for the remainder of the process. The model forced 100% of the melting to take place between the solid bed and the barrel wall. The model that we developed using screw rotation boundary conditions had about 80% of the melting occurring between the solid bed and the barrel wall and the remaining 20% between the solid bed and the screw core. The new melting model provides a realistic view of the complete melting process and not just for the first 25% as quantified by the Tadmor and Klein model. Solid bed breakup during melting has been an unsolved problem since it was first discovered by Maddock in 1959. This problem remained unsolved because previous researchers always viewed the problem with barrel rotation boundary conditions. The new melting model with screw rotation boundary conditions clearly showed that solid bed breakup is caused by the backwards transport of molten polymer by the screw core underneath the solid bed. This flow creates a relatively high pressure that can cause the breakup of the solid bed.

The impact of this discovery are numerous and include an improved fundamental understanding of the single-screw extrusion process and better models for predicting flows, pressure, and material temperature. This enables better simulation of processes and optimized designs for commercial extrusion processes. The research culminated with the publication of “Analyzing and Troubleshooting Single-Screw Extruders” that was coauthored with Mark A. Spalding. This is the first and only book that describes the single-screw extrusion process using the actual boundary conditions of screw rotation.