Response team investigates wildfire damage to buried drinking water infrastructure

Over the years, wildfires have crept out of the wild and into the urban landscape, threatening more lives, property, and infrastructure than ever before, sometimes devastating communities. In Paradise, California, and the surrounding Butte County, 85 people lost their lives and an estimated 18,800 structures were destroyed because of the November 2018 Camp Fire. Damage to aboveground infrastructure is easy to observe and assess. However, communities are only just beginning to understand how their buried drinking water infrastructure can be damaged or compromised during these disasters. A response team from Purdue University and Manhattan College led an investigation into the damage sustained to Paradise’s drinking water infrastructure and came to some surprising conclusions.

The Camp Fire, which began on Nov. 8, 2018, raged through Paradise, California, and the surrounding Butte County for more than two weeks, killing 85 people and destroying 18,804 structures. Not only was there damage to aboveground infrastructure, but belowground contamination of the area’s drinking water infrastructure was also a major concern. 

In early 2019, a response team composed of experts in various engineering disciplines from Purdue University and Manhattan College investigated the damage done to the Paradise Irrigation District’s drinking water infrastructure. Others eventually joined this effort, including collaborators from California State University, Chico; University of California, Berkeley; and Butte College. Purdue students were involved in the later phases of the response.

The Camp Fire physically damaged some of PID’s water infrastructure assets, including the reservoir cover, hydrants, meter boxes, and meters. These assets were bypassed, repaired, or replaced, as the utility generally understood what needed to be done — at least where the damage was visible. However, it was not clear how to recover from the contamination PID detected in its drinking water. The water utility determined that its water sources — which were above the fire line — and its treatment plant water were free of contamination.

It is well known that wildfires can contaminate natural water sources (for example, through ash deposition and soil runoff), and the solution is generally to modify treatment at the treatment plant to remove contamination. But if the contamination did not originate at the source, where had it come from?

PID decided to test its water in the distribution system because there had been another drinking water disaster in California the year before. The 2017 Tubbs Fire damaged 5,600 structures in and near Santa Rosa, California. Soon after the fire, a customer complained about water odor, which prompted Santa Rosa Water to use its in-house laboratory to conduct a suite of tests. The tests revealed the presence of volatile organic compounds (including benzene, which exceeded the state’s safe drinking water limit, as well as toluene), which prompted a do-not-drink/do-not-boil-water advisory that stayed in effect for 11 months for a select area of Santa Rosa Water’s distribution network. 

Testing outside this select area also indicated contamination. Benzene concentrations as high as 40,000 ppb were found during Santa Rosa Water’s recovery. For perspective, a level of 500 ppb is the threshold for a characteristic hazardous waste (or the threshold to be considered hazardous), according to the Resources Conservation and Recovery Act.

The fire damage was more widespread during the Camp Fire than the Tubbs Fire, and PID, along with state and federal officials, was concerned about a long recovery time. Other nearby water providers were also affected by the fire and found contamination in their distribution networks. While the Tubbs Fire recovery was eventually successful, there is no good playbook for how to recover from contamination originating within the distribution network. Regulatory compliance for safe drinking water has historically been enforced at the point of entry to the distribution system (in other words, at the treatment plant), but wildfire contamination originates in and travels through the distribution system. Many questions had to be answered to facilitate recovery; the team had to determine the type of contamination, its origin, and its movements as well as how it was detected. 

What was the contamination? 
Laboratory testing confirmed that the water was contaminated with a variable mixture of VOCs and semi-VOCs that included benzene, toluene, dichloromethane, and styrene; all these contaminants were present at levels above the state or federal safe drinking water limits. Of the VOCs that were tested for by the utilities impacted by the Camp Fire, benzene was a primary focus by the state. In California, benzene has a low maximum contaminant level of 1 ppb, compared to the federal limit of 5 ppb. However, even samples with no detectable levels of benzene were found to exceed other limits. For example, a sample from the Tubbs Fire without detectable benzene had dichloromethane levels of 41 ppb, which far exceeded the federal limit of 5 ppb.

When data from the Tubbs and Camp Fires were reviewed, no clear pattern of contamination was found when considering the data temporally, spatially, or from the standpoint of the compounds detected. For example, water sampled at one house might have undetectable benzene, while the next house might have 10 ppb or more. The water sampling was conducted over months while water was moving out of the system; it was not conducted to determine the cause of the contamination and was limited by personnel available. To date, it is unknown exactly which compounds pose the greatest health risk after a wildfire because limited work has been conducted in this area. 

However, there is enough known about these compounds to know that they can cause mild to serious risks to humans. Some of these compounds are harmful at lower concentrations than when they are detected by the human nose. Being exposed to high concentrations of these compounds for brief periods can cause adverse health effects like nausea, headaches, and fainting, whether through ingestion, inhalation, or dermal exposure (such as through personal hygiene activities or cooking). Acute exposure can cause neurological, immunologic, and hematologic effects. Chronic exposure has been linked to various cancers as well as reproductive and blood disorders. Because of their smaller mass, children may experience health effects at lower concentration exposures than adults. 

Given the multiple exposure routes, some utilities dealing with contamination have issued explicit showering directions such as taking lukewarm showers and ensuring that there is proper ventilation when these compounds are detected in the water supply. This prescribed shower temperature is based on the increased volatilization rate at higher temperatures rather than calculated chemical exposures based on models. 

Where did the contamination come from?
There are several hypotheses for how VOCs entered PID’s water system. Some of the compounds have been associated with smoke and ash, both of which may have been sucked into the system during the fire or during the recovery efforts. Additionally, PID’s staff observed negative pressure events due to leaks. Personnel will often work to close control valves and service lines to help maintain pressure and direct water resources where they are most needed. However, the Camp Fire came so quickly to Paradise that these precautionary efforts could not be completed at every service line. 

Thermal degradation of plastics installed within the distribution system and building plumbing may have also contributed to the water contamination. Plastics are ubiquitous in modern piping systems — even with metal pipes, plastics are used at joints, meters, or faucet connections. They are also increasingly popular; just before the fire, PID had updated a large portion of its metal system to polyvinyl chloride and high-density polyethylene pipes.

During recovery, PID found many melted and damaged plastic parts, including pipes that were completely blocked by another component’s melted remains. The plastics waste handling industry has shown that thermally treated plastics can release VOCs into the air. Tests by the Purdue team are ongoing to examine whether these same thermal processes enabled plastics to release chemicals into the water. 

In addition, plastic components could have been damaged within the distribution system — even underground, temperatures were elevated — or in homes and subsequently sucked back into the distribution system. 

How did the contamination move through the system? 
After a wildfire, a hydraulic model for a distribution network likely has little value because of the leaks and changed hydraulics in the system; the model is no longer calibrated. PID found and repaired leaks and main breaks in its system for months after the fire, especially as debris was moved for rebuilding efforts. Even after containment, fighting spot fires may direct water flow in odd directions. It is nearly impossible to gauge how water sloshes around a buried and damaged piping network after a fire, especially if precautions like valve closings were not taken. This sloshing may spread contamination throughout the system.

It is well known that some plastics can act as a sink for contamination like sponges, soaking up organic contamination. Later, when clean water moves through the infrastructure, this contamination can slowly leach back into the water. According to models developed before the fire by the U.S. Environmental Protection Agency, it could take as many as 195 days of constant flushing to reduce 20 ppb of benzene water contamination to less than 0.5 ppb from a single HDPE pipe. Moreover, replacing certain pipes may be more effective in terms of cost and time than flushing contamination out, as both the City of Santa Rosa and PID discovered.

Interestingly, the highest level of benzene contamination found after the Tubbs Fire (40,000 ppb benzene) was in a copper service line five months after the fire. Most service lines in the burn area were copper and hydraulic movement of contaminated water post-fire is unclear, so this is not entirely surprising. 

More work is needed to determine how biofilms, scales, and pipes interact with contamination caused by fire.

How was the contamination detected?
Water samples were taken in a variety of ways, including grab samples from hydrants, home fixtures, and water meters. A sampling device that connected to the water meter enabled easy and consistent water sampling for PID. To detect contamination from pipes that might be acting as sinks, a standard stagnation time of 72 hours was adopted by both Santa Rosa, within its advisory area, and PID. Clean water was flushed into the system and allowed to stagnate within the pipes for 72 hours, and then a water sample was collected. This 72-hour period was chosen as a manageable time that would also allow sufficient chemical leaching from the pipe (if a chemical were present). 

When grab samples were taken without preparation or measured stagnation — as was done by Paradise’s neighbor and Santa Rosa outside the advisory area — it could not be determined whether water had been sitting for two hours or two weeks. Thus, contamination may have gone undetected or been overestimated in comparison with other data. 

More research is needed to determine what kind of testing is most appropriate to identify health risks.

A typical testing platform for VOCs is gas chromatography-mass spectrometry. A laboratory that analyzes water samples will require that they arrive in headspace-free vials, which means no air or bubbles in the sealed bottle. If the sample comes into contact with air, the target compounds will volatilize or evaporate. Preservatives might also be added to a sample to quench the chlorine residual. The result is typically a chromatogram, which looks like an earthquake detection line, with spikes occurring at specific times for specific compounds. 

For each compound to be confirmed in water, a standard or pure form of that compound must be injected into the gas chromatograph. Benzene can appear quickly along the chromatograph, while other compounds can take longer to appear. Companies may offer benzene-only testing, which is less expensive than looking for the other compounds also present after wildfires. It can be challenging to determine which test is appropriate, but given the myriad compounds present and the lack of a clear pattern (for example, the presence or absence of benzene was not a predictor of other contamination), the response team recommended looking for a wide suite of compounds. More research is needed to determine what kind of testing is most appropriate to identify health risks. Additionally, looking for semi-VOCs requires a different test.

The path to recovery
When PID began planning its recovery, it looked to Santa Rosa’s recovery from the Tubbs Fire. Santa Rosa collected 5,244 water samples in the first eight months of recovery, including testing 3,103 service line locations that were within the advisory area or within 500 ft of a burned parcel. If contamination exceeded 0.5 ppb, which is less than the California maximum contaminant level of 1 ppb, or the home was within the advisory area, service lines were replaced. Some of the newly installed service lines became contaminated and had to be replaced again. Santa Rosa also replaced 2,875 water meters because of direct fire damage or contamination. A long-term water quality monitoring campaign was also put in place. A sample-then-replace-if-needed approach was chosen over a complete system rebuild because of cost considerations, many of which were recovered through the Federal Emergency Management Agency.

PID chose a similar path to recovery, but instead of 5 mi of its system being damaged, Paradise had 172 mi of distribution system damage. A do-not-use/do-not-boil mandate was issued for the water utility’s entire service area. Because people with standing homes wanted to move back in quickly, the utility began systematically clearing its distribution system. Once the utility returned pressure and did not find contamination in certain pipelines, it tested individual utility-owned service lines. Contaminated utility-owned service lines were replaced, and then advisories were lifted for each specific property. Property owners were responsible for testing their side of the water meter, the customer service line, and plumbing. While most advisories are now lifted, the utility recognized that property owners may not have removed all their contaminated infrastructure. 

PID chose a similar path to recovery, but instead of 5 mi of its system being damaged, Paradise had 172 mi of distribution system damage.

While Santa Rosa and Paradise used similar approaches, a water district neighboring Paradise never issued a do-not-use/do-not-boil order despite finding 530 ppb benzene. Some of this utility’s notices, which were endorsed by the state, indicated that odor was a good indicator for water contamination; however, it is not. Testing was offered by the utility in customer homes at a cost to the customer, and stagnation periods were either nonexistent or lasted eight hours rather than the 72 hours applied in Santa Rosa and Paradise. Different policies were applied to protecting the health of neighboring communities even for the same disaster.

A number of competing policy and financial limitations influenced drinking water contamination response and recovery decisions. In particular, technical support to commercial and residential building owners and occupants seemed to fall through the cracks. Utilities do not typically advise property owners on plumbing issues. The response team learned that many homeowners installed home drinking water treatment systems after the Camp Fire, and some chose to test their own water. Plumbing test methods changed during the response, and different organizations recommended different approaches. Ultimately, it was unclear whether plumbing was contaminated, and if it was, whether home treatment options were sufficiently protective.

The response team advised Butte County Public Health to require in-building sampling before allowing businesses to reopen, but it is unclear whether both hot and cold water, which have separate plumbing, were tested. Water quality can vary significantly spatially and temporally even within plumbing for the same building.

An ideal response
There is room for improvement in the way agencies and water utilities respond to wildfire-caused water contamination. One starting point is water advisories. While it is common to issue a boil-water advisory when pressure is lost (which can allow for bacterial contamination), a do-not-use order is most appropriate after wildfires. Since some of the contamination is volatile, boiling can speed the transfer of the chemicals from the water into the air, and the chemicals can then be inhaled. Bathing in contaminated water can pose increased health risks, especially to infants and children. Activities like toilet flushing and showering can also enable inhalation and dermal exposure. 

There is room for improvement in the way agencies and water utilities respond to wildfire-caused water contamination. One starting point is water advisories.

In a rapid community assistance survey implemented six months after the Camp Fire, the response team found that many people were still using water despite warnings that it could be unsafe. 

Response plans vary by disaster, but there are many questions utilities, public health departments, and local governments should ask themselves before disaster strikes. They include: 

• How will drinking water contamination be detected if it exists? What contaminants will be measured? 
• What equipment is needed for sampling? 
• Is it necessary to mobilize help from outside the disaster area? 
• Are local laboratories able to analyze samples? 
• What criteria will be used for asset replacement decisions? Who will be responsible for these costs, especially for the consumer’s side of the service line? 
• At what point is complete infrastructure replacement more cost-effective than alternatives like flushing or localized repairs? 
• What agencies need to be involved, and who will manage which aspects of the response? 
• What should residents in an impacted area do while the water systems are being investigated and repaired? 

Working through these questions before a disaster can help expedite response and recovery.

If water sampling is pursued, it is critical to develop an accurate and effective standard operating procedure quickly. To have comparable data that is likely to capture contamination, standard stagnation times should be used. This may require cooperation from customers and providing alternative water during the stagnation and testing period. Laboratories used should also be verified for accuracy. Periodic quality-control checks, such as sending duplicate samples to other labs, should be enforced. People who conduct testing will also need training on how to collect samples.

A recovery plan must also include education and instructions regarding plumbing. While water utility operators understand how the water flows through distribution systems and how contamination might spread, building owners do not have the same knowledge of their plumbing, which can be complicated, and the data are not yet available on how fire contamination might spread through the cold and hot water plumbing. A survey indicated that some people in the Camp Fire area relied on refrigerator granular-activated carbon filters for contamination removal. While such filters can remove some VOC contamination, simple calculations show that they could get saturated and thus become ineffective very quickly. Since current policy makes building owners responsible for finding and removing any contamination in their plumbing — even if it might originate from the distribution network — they need better guidance.

Design for disaster
While responses can clearly be optimized, only so much can be done with existing infrastructure. Design changes to the water system can further increase a community’s resiliency. Even simple modifications like integrating a sampling port near water meters designed for rapid water sampling could expedite a utility’s response to a wildfire. 

Materials should be carefully selected in fire-prone areas. For example, HDPE is more susceptible to becoming a sink for organic contamination than PVC. Thus, PVC service lines may be easier to properly flush in the event of contamination, though both are susceptible to heat damage. As scientists gain more information as to how plastics respond to fire damage, communities may be able to adopt codes and standards for fire-resistant distribution systems for fire-prone areas. In reality, numerous factors will affect material choices, including cost, other water quality issues like corrosion, and serviceability during extreme weather.

Researchers will continue to investigate new solutions, but practicing engineers, regulators, and citizens should work to implement specific policy changes now. 

Effective system isolation methods should also be considered. Designing accessible corporation stops, integrating more backflow prevention (preferably at every service line), and incorporating automatic shut-off valves could help isolate contamination within the system. Service line isolation could prevent contamination originating in plumbing from contaminating a distribution system and vice versa. In the rebuild after Camp Fire, backflow prevention was required on every new service line. 

If a shut-off valve can be automated, valuable utility work efforts could be spent on other response efforts such as ensuring that there is an adequate water supply to fight fires. Still, innovations will be needed to allow continued fire protection on individual properties.

Protecting the system from contamination may be possible with deeper burial or protective casing, but this goal needs to be balanced with accessibility for repairs, separation from other water systems, code requirements, and pipe durability to soil pressure. Moreover, many fire-prone areas overlap with earthquake-prone areas and thus compete for disaster-protection interests. Protecting plumbing from fire damage may be impossible.

Civil engineers have the power to adapt to this threat and protect drinking water systems and the communities that rely on those systems. As of 2019, there were only two well-documented cases of this type of contamination, and the initial discovery was due to a chance coincidence: the presence of compounds above high-odor thresholds, a customer perceiving this potential issue, and a utility with the resources to properly investigate. 

In the 2020 fire season, chemical contamination was found at more water systems. Contamination of drinking water by VOCs is likely much more widespread and should be considered in response to all wildfires that spread to urban areas. Water advisories should be proactive and preemptive rather than reactionary to avoid the confusion associated with rapidly changing advisories. Testing should be comprehensive, and ample flushing or asset replacement should be completed before people are allowed to use the water again. Appropriate hydraulic appurtenances and design changes can also better prepare infrastructure systems from this growing threat.

Researchers will continue to investigate new solutions, but practicing engineers, regulators, and citizens should work to implement specific policy changes now. 

This article first appeared in the January/February 2021 issue of Civil Engineering as “Fire and Water.”