Jeremy M. Gernand
Originally from the Houston, Texas area, I earned a B.S. in Mechanical Engineering from Texas A&M University. Following two years in the U.S. Peace Corps as a Math and Physics teacher in Guinea, West Africa, I returned to Houston as a Safety and Reliability Engineer at NASA′s Johnson Space Center. There I analyzed the severity and likelihood of the risks of engineered systems contributing to improvements in the safety and reliability of the International Space Station crew health care equipment. I later continued these efforts with greater quantitative emphasis at Northrop Grumman Electronic Systems as a reliability engineer predicting failure rates of new equipment.
During the same time, I earned a M.S. in Mechanical Engineering from Rice University conducting research into micro-scale fluid and heat transfer, and designing a microchannel methanol-steam reformer. Now, at Carnegie Mellon University pursuing a PhD in Engineering and Public Policy, I am working to develop and practice methods that enable data-driven quantitative risk assessment in the early stages of technology development.
Current Research Projects
Exploring the Factors Affecting Pulmonary Toxicity of Nanomaterials
Given the unique properties of nanomaterials (substances in solid form smaller than 100 nanometers) in which they can act both unlike bulk solids and also unlike chemical solutions, it is possible that new risks to human health or the environment may arise from a nano-version of a material we already know. Since we do not have the same wealth of experience with these materials as we do with other potential chemical pollutants, there is not yet agreement on which specific measurements or attributes of the nanomaterials will be necessary to predict risk.
This work has focused on a meta-analysis of in vivo pulmonary carbon nanotube exposure experiments with machine learning models that can effectively extract information from the sparse data set. So far, results have shown that carbon nanotubes with thicker diameters are more inflammatory, and that the catalytic metal impurities, especially cobalt, do display a sensitizing effect, but carbon nanotubes by themselves continue to contribute to toxicity.
▪ Journal article (Accepted Pending Minor Revisions, Risk Analysis), Author′s pdf
Further application of these same methodologies to data for nano-titanium dioxide particles appears to indicate that toxicity is increased will less aggregation and smaller particle sizes. These effects in pulmonary exposures seem to dwarf any positive or negative contribution from chemical factors such as crystalline structure, hydrophobicity, or purity. Overall, TiO2 toxicity seems to be attributable to deposited particulate mass and the length of the recovery period.
Quantitative Structure Activity Relationships or QSARs were originally envisioned as a way to predict the behavior of a new chemical (e.g. potential for bioaccumulation in the environment) from a set of specific measurable properties of that chemical such as molecular weight, polarization, and the octanol-water partitioning coefficient. The development of similar models for nanomaterials based on in vivo results would offer guidance to regulators and nanotechnology designers to reduce risks. While current data are too uncertain to produce this kind of mechanistic model, the random forest-based meta-analysis can offer guidance to creators of nanoparticles on tradeoffs between particle characteristics and their effects on toxicity.
▪ Conference paper (Accepted, Proc. IMECE2013), Author′s pdf
▪ Published machine learning models (via nanoHUB.org)
Setting Appropriate Occupational Exposure Limits for Nanomaterials
Currently proposed occupational exposure limits for nanomaterials like these reports from NIOSH on titanium dioxide and carbon nanotubes assume that every kind of nanoparticle made of those materials below the arbitrary 100 nm size category acts similarly. In fact, many nanomaterials display significant shifts in properties through this range, and a 15 nm particle may act very different in the human body from a 90 nm nanoparticle. My work in this area proposes a framework whereby safe exposure limits can be established for nanomaterials of a variety of sizes and shapes by using treed exponential models to generate a family of dose-response curves separated by the greatest toxicity-influencing factors.
Identifying Poor Reliability Parts from Logistics Warehouse Orders
As a Summer Research Associate with the RAND Corporation, I investigated the potential to identify poor reliability parts from changes in trends of parts orders recorded by a centralized logistics warehouse database. Accurately tracking the usage rate of a widely dispersed fleet after the end of the warranty period can be difficult or impossible, yet understanding that usage rate is critical to parsing out reliability issues from high rates of otherwise normal consumption.
Design of a Methanol Microreformer for Portable Fuel Cells
As fluids pass through smaller and smaller channels Newtonian behavior and common assumptions like the no-slip condition begin to break down, since molecules start to act more like beads running through a pipe rather than a continuum of infinitely differentiable units. My work involved the computationally intensive optimization of a channel geometry operating in the transition range between microchannel flow dynamics and Newtonian flow dynamics. The result was the design of a small scale device to produce hydrogen gas from methanol sufficient to power a 1 Watt fuel cell.
▪ US Patent (No. 8,034,134; Oct 2011)
▪ MS Thesis (Rice University)
The Health and Safety Criticality of Microgravity Countermeasures Equipment
While in a microgravity environment, the human body adapts in ways that can be detrimental to health and safety upon return to Earth. These effects can be slowed and sometimes halted by the use of countermeasures, mostly specialized exercise equipment. However, NASA initially classified these countermeasures as non-critical equipment given their seemingly simple designs, ease of repair, and lack of immediate failure consequences. Experience in the early stages of the ISS revealed that the non-critical designation adversely affected design for reliability, and that logistics supply issues often prevented timely resupply of spare parts contributing to a significant lack of system availability.
Other Research Interests
System Reliability Prediction, Monitoring, and Design
In practice, there is often inadequate feedback between quantitative reliability analysis of field performance and predictive reliability modeling or estimation based on design parameters or accelerated testing programs. The data needs of those managing or sustaining products in the field differ greatly from analysts and those designing new engineered systems. Technology now exists however that can provide the same kind of nearly instant feedback on system reliability that design engineers now receive for mechanical and thermal stresses. My work in this area seeks to apply automated learning algorithms to close the loop between field experience and system reliaiblity prediction and appropriately supply probabilistic risk information in near real time to design engineers as they create new systems.
Quantitative Learning, Uncertainty, Dirty Data, and Risk Assessment
Quantitative risk assessment in the early stages of the development of a new technology typically focuses on worst-case scenarios, expert elicitation, or broad trend projections based on past experience. Next, the wait sets in until sufficient data have been collected to satisfy statistical significance tests. Depending on the number of experimental factors involved this wait could be considerable. But now, with the advent of new machine learning algorithms and linked online databases, active risk management based on emerging data is becoming possible. My work in this area focuses on performing risk analysis with sparse data sets and developing metrics to evaluate tradeoffs between complex and traditional mathematical models.
Agricultural Systems and Nutrition
By many measures agriculture is the most dominant impact of human civilization on the global environment. Yet, this system oversupplies easily digestible calories and undersupplies vegetables in the U.S., while undersupplying adequate nutrition in the developing world. With world population now over 7 billion and heading upwards of 9 billion by 2050 combined with the uncertainty of climate change effects, the management of agriculture will be critically important to future human health and security. A systems optimization approach to this problem may provide useful criteria with which to benchmark our progress in achieving stability in our food supply.
Engineering and Public Policy
Baker Hall 129
Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh, PA 15213
PhD Candidate | Carnegie Mellon University
Engineering and Public Policy
jgernand [at] andrew [dot] cmu [dot] edu
Resume and CV
CMU | Carnegie Mellon University
EPP | Engineering and Public Policy Department
CEINT | Center for the Environmental Implications of Nanotechnology
RAND | RAND Corporation
JHSPH | Johns Hopkins School of Public Health
NGES | Northrop Grumman Electronic Systems
RICE | Rice University
SAIC | Science Applications International Corporation
JSC | NASA Johnson Space Center
USPC | U.S. Peace Corps
TAMU | Texas A&M University
Contact Me [pdf]
Doctoral Dissertation Defense for Jeremy M. Gernand
10:00 am, Friday, June 28th, 2013
Baker Hall 129