A new study from the Environmental Defense Fund indicates that 72% of methane emissions in the Permian Basin come from small sources (leaks rather than major ruptures or releases) <https://apnews.com/article/methane-gas-texas-permian-basin-satellite-climate-81b9cfea311275806c6fbf8621f821c2>. This is a sign of progress, though it means further improvements will be more labor intensive. Methane has significant climate change impacts, and while not toxic itself, it can lead to increased human exposures to ozone and other toxic chemicals created due to reactions in the atmosphere. Reducing methane emissions will have concrete health benefits to people who live nearby.

My group is currently working on a study of Marcellus area air emissions including methane, NOx, SOx, and PM. While the number of wells is less than in the Permian, this remains a significant source of regional air emissions. We hope to publish results later in 2025.

The Federal Rail Administration (FRA) just announced the finalization of a new regulation to improve the job safety of rail transit workers in response to dangers from assaults. Over the past 16 years at least 29 transit workers have been killed and 144 injured due to assaults (https://www.transit.dot.gov/about/news/biden-harris-administration-announces-new-rule-protect-rail-transit-workers-nationwide).

“On average, the rule would prevent an estimated 1.4 fatalities and 3.9 injuries per year, resulting in annual safety benefits of $14.2 million in 2021 dollars. To meet the safety standards, [transit agencies] and [state safety oversight agencies] would incur an estimated $2.0 million in start-up costs plus $11.3 million in ongoing annual costs.” (https://www.federalregister.gov/documents/2024/03/25/2024-06251/rail-transit-roadway-worker-protection)

The House Space and Aeronautics subcommittee recently held a hearing on commercial space utilization and public safety (https://science.house.gov/2024/9/space-aeronautics-subcommittee-hearing-risks-and-rewards-encouraging-commercial-space-innovation-while-maintaining-public-safety). A key issue at this hearing was the tension between preserving public safety as a priority and also facilitating innovation and more rapid changes. You can read the chair and witness statements at the link above. The increase in the pace of commercial space launches and overall utilization puts pressure on the existing regulatory apparatus to appropriately review these operations, and it certainly seems like increased investment whether funded through general appropriations or through increased fees on the part of the launch service companies would be appropriate to not compromise the historically high safety standards in this industry.

I had a great time presenting a lecture on the application of CART machine learning models to risk analysis at the Civil Aviation University of China yesterday. Many thanks to my hosts. I look forward to further opportunities for collaboration in the applications of advanced data analytics to questions on risk analysis and management.

CART models have several valuable characteristics for risk analysis including learning more effectively from incomplete data, identifying critical thresholds, identifying conditionally important variables, and interpretability. I have employed CART models and random forest models in a variety of risk-related applications from mine worker safety to nanoparticle toxicity, and always looking for more that they can do.

A thing that doesn’t get talked about enough is the accompanying benefits of reducing CO2 emissions. This new paper does that and puts a dollar value on those benefits: https://doi.org/10.1016/j.egycc.2024.100165.

When we stop emitting CO2 to the air, we also stop emitting a lot of other pollutants that increase people’s risks of cardiovascular disease and cancer and other problems. This analysis estimates $65 to $250 billion in annual benefits from this reduction in non-CO2 emissions, for only partly achieving zero emissions goals. That’s big.

While it feels nice after the end of Daylight Saving Time (DST) in the Fall to wake up or head into work or school with the sun and not in darkness, changes into and out of DST are both associated with increases in vehicle and other forms of home and work accidents (https://doi.org/10.1016/j.aap.2017.11.029) (https://doi.org/10.1016/j.aap.2017.11.029).

The reasons for DST are no longer pertinent in our internet-connected world and this applies to time zones as well. We can make our own local decisions on school start times and business hours and make national and international coordination easier. Permanent DST (or ending DST) is just a necessary start, though one we should implement immediately. The human costs of these changes are unnecessary, and we owe it to each other to end them.

The biggest risk to kids on Halloween comes from collisions with automobiles. With all the excitement of friends about in the neighborhood after dark and wearing special costumes, it can be difficult for cars to see children and challenging for kids to see and recognize the danger from cars. Adults need to take special care to supervise street crossings even for kids who are “too old” to normally need such supervision. Finding ways to make costumes reflective, incorporating light features into costumes, and taking action as a neighborhood to slow local traffic on this one evening are valuable mitigations that can help prevent serious injuries. Being safe is more fun! <https://www.thefloridalawgroup.com/news-resources/5-halloween-accident-statistics-you-should-know-5-halloween-safety-tips-for-children-parents-this-weekend/>

In this undergraduate-led research project that we concluded in 2020, we found small quantities of nanoparticles in the breathing zone during application of aerosol mineral sunscreens recommended for use on children (https://doi.org/10.1007/s41810-020-00079-x). Overall, 85% of the particles by count had diameters less than 100 nm, and this project captured direct images of several different particle types with some clearly the product of mineral grinding processes, while others show signs of having been manufactured. Interestingly, plume sizes for the 3 tested products were widely different with some extending more than 2.5 m from the pressurized can. There seems to be little risk under the particular scenario we tested, as the highest measured concentration for all particles was 0.8 micrograms per cubic meter, and the lowest recommended exposure limit for titanium dioxide nanoparticles (from NIOSH) is 300 micrograms per cubic meter.

Hawaii and other states and localities are in the midst of considering new rules and laws about the operation of electric bikes (ebikes) [https://www.hawaiinewsnow.com/2024/10/23/officials-brainstorm-new-rules-e-bikes-improve-safety-amid-rising-accidents/]. As an ebike rider myself, I am both encouraged by and a little leery of new regulations for ebikes. Many jurisdictions, however, are stuck with rules developed for gasoline powered mopeds more than 40 years ago. Ebike technology is a potentially excellent way to increase mobility and reduce car use at the same time by making it easier to make trips on bicycle regardless of physical fitness. But ebikes are heavy, and children and others can easily get to speeds and situations that are beyond their experience and capabilities and put themselves at risk. Prioritizing facilitating this transportation method, while keeping riders away from cars first, and separated from pedestrians second, can help our towns and cities get the most out of this technology.

A recent article in Scientific Reports (https://doi.org/10.1038/s41598-024-74345-0)(https://www.azom.com/news.aspx?newsID=63822) demonstrated the efficacy of copper and zinc oxide nanoparticles in reducing corrosion. Corrosion is a major problem worldwide that requires large amount of labor and resources to combat and limits the useful lifetime of metal components that are energy intensive to manufacture. As we explore benefits like this, we need to keep in mind the potential environmental and occupational health risks and take action now to limit the potential downsides of this technology. As these particles are manufactured and processed into coatings and applied to products, we need to keep them from washing away into the environment and from being inhaled by workers and maintainers.

My student Vignesh Ramchandran and I made an effort in this 2023 paper to estimate the NOAEL (No Observed Adverse Effect Level) for groups of engineered nanoparticles (https://doi.org/10.1115/1.4055157). The NOAEL is the highest dose of an exposure that produces no detectable negative effect in the subject (usually rodents). It is the basis for setting safe exposure limits for people, where it is usually reduced by a factor of between 10 and 100 due to humans’ longer lifespans and other factors.

Interestingly, we identified clusters of nanoparticles that displayed similar combinations of effects with a regression tree model, and estimated NOAELs between 20 and 16,000 micrograms/kilogram for 8 clusters of metal oxide nanoparticles. Uncertainty remains an issue given the data available at the time, but seeing several orders of magnitude difference from one cluster to the next suggests that we might not want to treat all these nano-substances as the same thing when creating regulatory limits.

The Supreme Court recently declined to review and potentially invalidate the leadership structure of the Consumer Product Safety Commission (CPSC) (https://thehill.com/homenews/4944532-supreme-court-consumer-safety-agency/). The CPSC is led by 5 board members, political appointees confirmed by the senate, who serve 7-year terms and can only be removed for cause (rather than at the president’s discretion). Some safety regulators have similar structures like the Nuclear Regulatory Commission (NRC), whose members serve 5-year terms, while many have single directors or administrators like OSHA, MSHA, the EPA, and the FRA. The Supreme Court did not elaborate on why they declined the case, so it is possible an entity directly regulated by one of these agencies could bring a similar challenge in the future.

I have never seen a comprehensive review or analysis on the effectiveness of regulators lead by a board versus a single administrator, so I wonder what the difference is in reality. Please point me to one, if you know. Proposing a board rather than a single administrator is often a compromise measure for Congress to moderate the management of an agency that will initially be overseen by a president of an opposing party. The laws that govern rule making, however, are the same.

The safety review of Boeing by the FAA (https://www.reuters.com/business/aerospace-defense/faa-says-new-boeing-safety-review-will-take-about-three-months-2024-10-18/) is needed to restore confidence by the public and corporate customers and ensure continuation of commercial aviation in the US. Safety culture is a tenuous thing and is challenging to hold on to during times of restructuring of the kind Boeing has undertaken in the recent decade (having suppliers take on more of the design and assembly tasks than before). But safety reviews or audits only work when taken seriously by both sides and when the audience for that review includes all stakeholders like customers. As uncomfortable as it can be for open discussion of failings, this is a critical part of getting risk management right.

This recent opinion piece states that existing US radiation exposure limits are too strict and that this has led to excessive fear of radiation and a lack of investment in US nuclear power (https://www.realclearenergy.org/articles/2024/10/14/radiation_time_forgotten_1065066.html).

The article does correctly point out that many of our modern exposure standards are set based on what we learned from the single one-time dose Japanese residents received during the atomic bomb explosions at the end of World War II. Environmental health experts have struggled since then to make sense of that event and its health effects with regard to more mundane chronic exposures like radon, x-rays, and work-related exposures. More modern research can only find the level of exposure below which we lose the potential effect of radiation in the noise of other exposures, not really identify a safe threshold.

If it is true that there is no safe threshold of radiation exposure, or that the safe threshold is somewhat at or around average modern exposures, we may never be able to know or prove that to be true as it would require tracking lifetime exposures and health for tens or hundreds of millions of people, a significant practical challenge.

Bottom line: modern nuclear power is safe and reliable, and there is an ample safety factor built into current exposure limits, but the exposure limits are not the reason the public is wary of nuclear energy. We can do a better job of putting risks of other energy sources in context and hold on to nuclear power’s admirable safety record in the US.

In this study published in 2023 (https://asmedigitalcollection.asme.org/IMECE/proceedings/IMECE2022/86717/V009T14A037/1157514), we ran an experiment with 141 undergraduate engineering students and asked them to allocate a limited budget on a transportation infrastructure project between competing objectives of safety and profitability. It was possible to meet minimum requirements in both objectives, but not the ultimate goals for both objectives with the available budget, so we got to see where they put their “extra” money. We primed different groups with different management priorities, and participants didn’t always follow the stated priming (to maximize safety or profitability or achieve balance), but did report following their perception of the “manager’s” priorities (rather than claiming they relied on their own judgment). We’re thinking about revising this experiment and recruiting a larger sample size to see if we may be able to learn more.

The EPA is currently in the process of reducing the lead in drinking water action level from 0.015 mg/L to 0.010 mg/L (https://www.regulations.gov/document/EPA-HQ-OW-2022-0801-0036). Nearly 200,000 comments were received on this proposed rule, which will also mandate removal of many lead drinking water pipes in the US, and limit copper concentrations. Unlike some toxic substances, we don’t know of any safe level of lead exposure. If one exists at some extremely small value, we don’t have any information to know where that level would be. Our current risk models work in a way to predict that any amount of exposure could produce some harm, if a large enough group of people were affected. And drinking water standards affect most Americans through water we get at home, school, work, and other places. The reason why we allow some lead in drinking water is because it would be very costly to remove it all, and at some point, the risk is small enough that other priorities take precedence.

Reducing lead exposure has likely benefits in increasing health and IQ of children among other benefits like reducing some forms of cardiovascular disease and reducing instances of low birth weight. The total net monetary benefits of the rule are expected to be between $15.3 billion and $31.9 billion over the next decade (including the $5-8 billion in increased earnings from better IQs). Even though there are thousands of water systems in the US, controlling exposure through their regulation is a fairly efficient way to do so, as opposed to cost and reliability issues with trying to manage this by way of residential and business water filters.

This recent paper (https://doi.org/10.3390/app14167192) looked at whether people’s decision making in identifying real or fake (artificially generated) images improved when given suggestions from an AI. An accurate and less inaccurate version of AI were tested in the experiment. The paper found that if the AI’s skill level was better than the human participant’s, the participant’s performance improved similarly to how talking over a situation with a more experienced or more skilled colleague would help make a better decision.

This is perhaps unsurprising, but it does mean that people in this experiment were able to sense the AI’s skill level in relation to their own to some extent, and not go along when they knew better. In tasks related to risk or safety hazard identification, if this effect holds up, it could mean that AI helpers could enhance performance and make it less likely that potential risks or hazards would not be identified. Like many of these studies, the sample size was small (N=30), so further investigation is recommended.

This NPR story (https://www.kcur.org/news/2024-10-14/volunteer-fire-departments-worry-about-proposed-safety-rules) highlights a particular issue that often arises in regulatory scenarios: costs of a new regulation can be disproportionately high for small organizations like volunteer fire departments or small companies. When we are talking about small companies within a larger industry, we either exempt small businesses from the regulation, or we decide to live with the consequences of putting some companies out of business because the overall benefits of the regulation are very large in comparison to the costs to the industry as a whole.

OSHA found that this rule had $2.6 billion in annual benefits (54 fewer fatalities per year and numerous fewer cases of cancer and injuries) and $545 million in annual costs, which means that somewhere in the United States as a whole there is enough benefit to cover the cost, but those benefits may not be in the hands of these volunteer fire departments. If that is the case, and these values hold up to scrutiny and justify the new rule, governments should find a way to offset these costs, so that they can comply and reap the benefits along with the better funded professional fire departments.

Finally, a pet peeve… The proposed regulation together with its justification including all of the safety, health, and economic analysis ran to 608 pages with the required formatting of the federal register (the length mentioned in the article as a critique of its complexity), this is not the text of the rule itself, which is much shorter and concise. You can see this justification document for yourself here (https://www.federalregister.gov/documents/2024/02/05/2023-28203/emergency-response-standard?utm_campaign=subscription+mailing+list&utm_medium=email&utm_source=federalregister.gov).

I was curious to read this recent preprint on arXiv (https://doi.org/10.48550/arXiv.2410.02820) that describes the results of an experimental investigation into whether the GPT-4o model is risk averse and biased in a similar way to human subjects. The authors write that “GPT-4o consistently navigates conjunction fallacy and certain aspects of probability neglect with statistically robust responses. However, the model frequently falls prey to biases such as the resemblance heuristic and framing effects.”

So, the model does better than an untrained human in certain situations and similarly in others. While this experiment is limited to the kinds of language processing that GPT-4o should be able to handle, these models can be employed to give advice, and being aware of and developing methods to mitigate bias in these models will be an important area of research going forward.

If anyone is interested in setting up a GPT-like experiment similar to some of our previous engineering student experiments on risk-related decisions (like https://doi.org/10.1115/IMECE2022-95484 or https://doi.org/10.1115/1.4055156), let me know. I would be interested in running some head-to-head comparisons.

From this paper (https://doi.org/10.1016/j.jsr.2022.05.005), published in 2022, our analysis found that the California mandate to put solar PV systems on new residential rooftops was likely to modestly increase worker injuries in the state (residential rooftops are one of the most dangerous working environments in the US). The paper does not consider any offsetting health benefits from emissions-free energy production, but it does outline several alternatives to still obtain those benefits and increase solar PV energy production without the worker safety drawbacks (such as installing the solar panels at ground level nearby, rather than on rooftops).