Safety Standard for Portable Generators

Federal RegisterNov 21, 2016

Ask Donna

What actually matters in this document.

Text

CONSUMER PRODUCT SAFETY COMMISSION

16 CFR Part 1241

[Docket No. CPSC-2006-0057]

RIN 3041-AC36

Safety Standard for Portable Generators

AGENCY:

Consumer Product Safety Commission.

ACTION:

Notice of proposed rulemaking.

SUMMARY:

The U.S. Consumer Product Safety Commission has determined preliminarily that there may be an unreasonable risk of injury and death associated with portable generators. To address this risk, the Commission proposes a rule that limits CO emissions from operating portable generators. Specifically, the proposed rule would require that portable generators powered by handheld spark-ignition (SI) engines and Class I SI engines not exceed a weighted CO emission rate of 75 grams per hour (g/hr); generators powered by one-cylinder, Class II SI engines must not exceed a weighted CO emission rate of 150 g/h; and generators powered by Class II SI engines with two cylinders must not exceed a weighted emission rate of 300 g/h.

DATES:

Submit comments by February 6, 2017.

ADDRESSES:

You may submit comments, identified by Docket No. CPSC-2006-0057, by any of the following methods:

Electronic Submissions:

Submit electronic comments to the Federal eRulemaking Portal at:

http://www.regulations.gov.

Follow the instructions for submitting comments. The Commission does not accept comments submitted by electronic mail (email), except through

www.regulations.gov.

The Commission encourages you to submit electronic comments by using the Federal eRulemaking Portal, as described above.

Written Submissions:

Submit written submissions by mail/hand delivery/courier to: Office of the Secretary, Consumer Product Safety Commission, Room 820, 4330 East West Highway, Bethesda, MD 20814; telephone (301) 504-7923.

Instructions:

All submissions received must include the agency name and docket number for this notice. All comments received may be posted without change, including any personal identifiers, contact information, or other personal information provided, to:

http://www.regulations.gov.

Do not submit confidential business information, trade secret information, or other sensitive or protected information that you do not want to be available to the public. If furnished at all, such information should be submitted in writing.

Docket:

For access to the docket to read background documents or comments received, go to:

http://www.regulations.gov,

and insert the docket number CPSC-2006-0057, into the “Search” box, and follow the prompts.

FOR FURTHER INFORMATION CONTACT:

Janet Buyer, Project Manager, Directorate for Engineering Sciences, Consumer Product Safety Commission, 5 Research Place, Rockville, MD 20850; telephone: 301-987-2293; email:

jbuyer@cpsc.gov.

SUPPLEMENTARY INFORMATION:

I. Background

A portable generator is an engine-driven machine that converts chemical energy from the fuel powering the engine into rotational energy, which, in turn, is converted to electrical power. Reports of portable generator-related fatalities and injuries prompted the U.S. Consumer Product Safety Commission (Commission or CPSC) to publish an advance notice of proposed rulemaking (ANPR) in December 2006 to consider whether there may be an unreasonable risk of injury and death associated with portable generators (71 FR 74472 (December 12, 2006)). The ANPR began a rulemaking proceeding under the Consumer Product Safety Act (CPSA). The Commission received 10 comments in response to the ANPR. Subsequently, in a two-part technology demonstration program, CPSC contracted with the University of Alabama (UA) to conduct a low CO emission prototype generator technology development and durability demonstration and contracted with NIST to conduct comparative testing of an unmodified carbureted generator and prototype generators in an attached garage of a test house facility. CPSC staff published a report regarding the results of the UA technology demonstration and received 12 comments in response to this report. NIST published a report concerning its comparative testing of generators and received four comments in response to its report. The Commission is now issuing a notice of proposed rulemaking (NPR) that would establish requirements for carbon monoxide emission rates.

The information discussed in this preamble is derived from CPSC staff's briefing package for the NPR, which is available on CPSC's Web site at:

https://www.cpsc.gov/s3fs-public/ProposedRuleSafetyStandardforPortableGenerators.pdf.

The Commission voted (4-1) to publish this notice in the

Federal Register

. Chairman Elliot F. Kaye and Commissioners Robert S. Adler, Joseph P. Mohorovic, and Marietta S. Robinson voted to approve publication of the proposed rule. Commissioner Ann Marie Buerkle voted against publication of the proposed rule.

II. Statutory Authority

Portable generators are “consumer products” that can be regulated by the Commission under the authority of the CPSA.

See

15 U.S.C. 2052(a). Section 7 of the CPSA authorizes the Commission to promulgate a mandatory consumer product safety standard that sets forth certain performance requirements for a consumer product or that sets forth certain requirements that a product be marked or accompanied by clear and adequate warnings or instructions. A performance, warning, or instruction standard must be reasonably necessary to prevent or reduce an unreasonable risk or injury.

Id.

Section 9 of the CPSA specifies the procedure that the Commission must follow to issue a consumer product safety standard under section 7. In accordance with section 9, the Commission may commence rulemaking by issuing an ANPR; as noted previously, the Commission issued an ANPR on portable generators in December 2006. (71 FR 74472 (December 12, 2006)). Section 9 authorizes the Commission to issue an NPR including the proposed rule and a preliminary regulatory analysis, in accordance with section 9(c) of the CPSA and request comments regarding the risk of injury identified by the Commission, the regulatory alternatives being considered, and other possible alternatives for addressing the risk.

Id.

2058(c). Next, the Commission will consider the comments received in response to the proposed rule and decide whether to issue a final rule, along with a final regulatory analysis.

Id.

2058(c)-(f). The Commission also will provide an opportunity for interested persons to make oral presentations of the data, views, or arguments, in accordance with section 9(d)(2) of the CPSA.

Id.

2058(d)(2).

According to section 9(f)(1) of the CPSA, before promulgating a consumer product safety rule, the Commission must consider, and make appropriate findings to be included in the rule, on the following issues:

• The degree and nature of the risk of injury that the rule is designed to eliminate or reduce;

• the approximate number of consumer products subject to the rule;

• the need of the public for the products subject to the rule and the

probable effect the rule will have on utility, cost, or availability of such products; and

• the means to achieve the objective of the rule while minimizing adverse effects on competition, manufacturing, and commercial practices.

Id.

2058(f)(1). Under section 9(f)(3) of the CPSA, to issue a final rule, the Commission must find that the rule is “reasonably necessary to eliminate or reduce an unreasonable risk of injury associated with such product” and that issuing the rule is in the public interest.

Id.

2058(f)(3)(A)&(B). Additionally, if a voluntary standard addressing the risk of injury has been adopted and implemented, the Commission must find that:

• the voluntary standard is not likely to eliminate or adequately reduce the risk of injury, or that

• substantial compliance with the voluntary standard is unlikely.

Id.

2058(f)(3)(D). The Commission also must find that expected benefits of the rule bear a reasonable relationship to its costs and that the rule imposes the least burdensome requirements that would adequately reduce the risk of injury.

Id.

2058(f)(3)(E)&(F).

III. The Product

A portable generator is an engine-driven machine that converts chemical energy from the fuel powering the engine to mechanical energy, which, in turn, is converted to electrical power. The engine can be fueled by gasoline, liquid propane, or diesel fuel.

A portable generator has a receptacle panel for connecting appliances or other electrical loads

via a cord with a plug connection. Portable generators are designed to be carried, pulled, or pushed by a person.

Engines that operate on gasoline or liquid propane are called spark ignition (SI) engines and engines that operate on diesel fuel are called compression ignition (CI) engines.

An electrical load is an electrical component or portion of a circuit that consumes electric power. This is opposed to a power source, which produces power, such as a battery or generator. Examples of loads include: Appliances, lights, and power tools.

Portable generators that are the subject of the proposed standard commonly are purchased by household consumers to provide electrical power during emergencies (

e.g.,

power outages caused by storms), during other times when electrical power to the home has been shut off, when power is needed at locations around the home without access to electricity, and for recreational activities (

e.g.,

camping or recreational vehicle trips). Built-in wheels or optional wheel kits are often available for heavier, more powerful units (

e.g.,

units with 3 kW power ratings and more).

One of the primary features of a generator is the amount of electrical power the generator can provide on a continuous basis. This power, commonly referred to in the industry as “rated power,” is advertised in units of watts or kilowatts (kW), and can range anywhere from under 1 kW for the smallest portable generators, to nominally 15 kW for the largest portable generators.

Knowing the generator’s rated power is useful in choosing the appropriate size generator for a particular electrical load, such as providing power to power tools, household appliances, or recreational equipment.

As we will discuss further herein, the generator’s rated power is generally a function of the size of the engine. However, there is no industry standard for relating the generator’s rated power to the size of the engine; nor is there any uniform way in which electrical output capacity is advertised as “rated.”

IV. Risk of Injury

A. Description of Hazard

Carbon monoxide is a colorless, odorless, poisonous gas formed during incomplete combustion of fossil fuels, such as the fuels used in engines that power portable generators. The initial effects of CO poisoning result primarily from oxygen deprivation (hypoxia) due to compromised uptake, transport, and delivery of oxygen to cells. Carbon monoxide has a 250-fold higher affinity for hemoglobin than does oxygen. Thus, inhaled CO rapidly enters the bloodstream and effectively displaces oxygen from red blood cells, resulting in the formation of carboxyhemoglobin (COHb).

The heart, brain, and exercising muscle are the tissues with the highest oxygen requirements; consequently, they are most sensitive to CO-induced hypoxia. The CO-induced hypoxia is reflected in the non-specific, flu-like symptoms of mild CO poisoning and early symptoms of severe poisoning,

e.g.,

headache, lightheadedness, nausea, and fatigue. More severe CO poisoning can result in progressively worsening symptoms of vomiting, confusion, loss of consciousness, coma, and ultimately, death. The high CO emission rate of current portable generators can result in situations where the COHb levels of exposed individuals rise suddenly and steeply, causing people to experience rapid onset of confusion, loss of muscular coordination, and loss of consciousness. This can occur without people first experiencing milder CO poisoning symptoms associated with a low, or slowly rising, CO level.

COHb, expressed as a percentage, reflects the percentage share of the body's total hemoglobin pool occupied by CO. Although the relationship is not absolute, percent COHb levels can provide a useful index of CO poisoning severity. It is measured with a blood sample from the exposed person.

B. Incident Data

1. Portable Generator Carbon Monoxide Fatalities

The Commission publishes an annual report that summarizes CO incidents associated with engine-driven generators and other engine-driven tools.

The Commission is using this report to provide the base number of incidents for the rulemaking. CPSC staff set a date of May 21, 2015, as a cut-off for the incident data used in the briefing package. As of May 21, 2015, CPSC databases contained reports of at least 751 generator-related consumer CO poisoning deaths resulting from 562 incidents that occurred from 2004 through 2014.

Due to incident reporting delays, statistics for the two most recent years, 2013 and 2014, are incomplete because data collection is ongoing. Therefore, the numbers for these years will likely increase.

Figure 1 shows the count of deaths involving a generator derived from CPSC databases for each of these years. Note that reporting of generator-related deaths is not a statistical sample or a complete count of incidents.

These numbers are taken from a June 2015 reported by the CPSC, Hnatov, Matthew,

Incidents, Deaths, and In-Depth Investigations Associated with Non-Fire Carbon Monoxide from Engine-Driven Generators and Other Engine-Driven Tools, 2004-2014,

U.S. Consumer Product Safety Commission, Bethesda, MD, June 2015. (Docket Identification CPSC-2006-0057-0026, available online at:

www.regulations.gov

Id.

Note that the epidemiological benefits analysis and preliminary regulatory analysis, discussed in Sections IV and X, do not include the 85 deaths reported to CPSC as of May 21, 2015, for the years 2013 and 2014 because reporting for these years is considered incomplete. The epidemiological benefits analysis and preliminary regulatory analysis also exclude incidents involving generators that are out of the scope of the proposed rule (7 deaths in 5 incidents). Therefore, the Commission's epidemiological and regulatory analyses are based on 659 deaths in 493 incidents that occurred from 2004 through 2012.

EP21NO16.003

2. Portable Generator Carbon Monoxide Injuries

Based on CPSC's National Electronic Injury Surveillance System (NEISS) database,

CPSC estimates that for the 9-year period of 2004 through 2012, there were 8,703 CO injuries associated with generators seen in emergency departments (ED). This estimate should not be considered definitive because physicians have noted difficulty in correctly diagnosing these injuries. Carbon monoxide poisoning may mimic many nonfatal conditions, including alcohol or drug intoxication, psychiatric disorders, flulike illnesses, and other conditions that can lead to misdiagnosis. Measurement of COHb levels in the victim's blood, which could confirm CO poisoning, can also be confounded based on the time elapsed and any supplemental oxygen treatment administered, which can lower COHb counts prior to measurement. In addition, in some incidents, first responders transported severely poisoned victims found at the scene directly to a medical facility with a hyperbaric oxygen (HBO) chamber

for treatment rather than to a hospital ED. These incidents would not have been captured in NEISS. For these reasons, the Commission believes that the injury estimate for this proposed rule may be low.

The NEISS database is a national probability sample of hospitals in the United States and its territories. Patient information is collected from each NEISS hospital for every emergency visit involving an injury associated with consumer products. From this sample, the total number of product-related injuries treated in hospital emergency rooms nationwide can be estimated.

An HBO chamber is a facility used for exposing patients to 100 percent oxygen under supra-atmospheric conditions, to shorten the time it would otherwise normally take for the CO to leave the bloodstream and to increase the amount of oxygen dissolved in the blood. A broad set of recommendations has been established for HBO treatment for CO poisoning, which includes a COHb level above 25 percent, loss of consciousness, severe metabolic acidosis, victims with symptoms such as persistent chest pain or altered mental status, and pregnant women. Treatment is not recommended for mild-to-moderate CO poisoning victims, other than those at risk for adverse outcomes.

In addition to using the NEISS database to estimate CO poisoning injuries for the years 2004 through 2012, the Commission examined the narratives of the 292 records of CO-related ED visits to NEISS-member hospitals associated with generators for the years 2004 through 2014. The narratives helped illustrate the range of treatments received, the symptoms, and the reasons why victims went to a hospital ED.

Hnatov, Matthew,

Summary of NEISS Records Associated with Carbon Monoxide Exposure Cases Related to Engine-Driven Generators in 2004 through 2014,

U.S. Consumer Product Safety Commission, Bethesda, MD, November 2015. (Docket Identification CPSC-2006-0057-0028, available online at:

www.regulations.gov

The Commission used the Injury Cost Model (ICM) to estimate the number of injuries treated in locations other than hospital EDs. The ICM uses empirical relationships between the characteristics of injuries and victims in cases initially treated in hospital EDs and those initially treated in other medical settings (

e.g.,

physicians' offices, ambulatory care centers, emergency medical clinics), based primarily on data from the Medical Expenditure Panel Survey,

to estimate the number of medically attended

injuries treated outside of hospital EDs. The ICM also analyzes data from the Nationwide Inpatient Sample of the Healthcare Cost and Utilization Project

to project the number of direct hospital admissions bypassing the hospital EDs. According to the ICM estimates, there were an additional 16,660 medically attended CO injuries involving generators during 2004-2012. Consequently, based on NEISS and ICM estimates, there was a minimum of about 25,400 medically attended CO injuries treated during the 9-year period. This is a ratio of almost 39 generator-related CO injuries to every CO death that occurred in that period.

The Medical Expenditure Panel Survey (MEPS) is a nationally representative survey of the civilian non-institutionalized population that quantifies individuals' use of health services and corresponding medical expenditures. The MEPS is administered by the Agency for Healthcare Research and Quality (U.S. Department of Health & Human Services). The MEPS has been collected continuously since 1999 and is the principal data set used to monitor medical spending in the United States.

The National (Nationwide) Inpatient Sample (NIS) is part of a family of databases and software tools developed for the Healthcare Cost and Utilization Project (HCUP). The NIS is the largest publicly available all-payer inpatient health care database in the United States, yielding national estimates of hospital inpatient stays. HCUP is a family of health care databases and related software tools and products developed through a federal-state-industry partnership and sponsored by the Agency for Healthcare Research and Quality (U.S. Department of Health & Human Services).

Table 1 presents a list of the most commonly identified symptoms given in the NEISS case narratives of 292 cases involving generator-related CO injuries that occurred in the 11-year period from 2004 through 2014. In many cases, multiple symptoms were reported, but in 29 percent of the cases (85 of 292), symptoms were not described in the NEISS narrative, although the diagnosis was reported. The weighted proportion of the total appears to account for the selection probabilities of each case.

Table 1—Most Common Symptoms Reported in NEISS CO Poisoning or CO Exposure Cases Involving Generators, 2004-2014

Common symptoms *

Cases

Weighted

proportion

(%)

Headache

Nausea, Felt Sick

Dizzy/Confused, Disorientation, Lightheaded

Vomiting

Passed Out, Unconscious, Unresponsive

* Cases may appear multiple times in Table 1 because victims may have exhibited multiple symptoms.

Table 2 presents a summary of the reasons why the patients said they went to the emergency room for treatment or to be checked out. In the majority of cases, the medical records, from which the narratives were abstracted, provided little or no information on how the patients knew they needed to go to the emergency room or how they got there. However, in 47 of the 93 cases in which this information was available, the patient realized something was wrong and arranged to get to the emergency room.

Table 2—Reason Victim Went to ED for NEISS CO Poisoning or CO Exposure Cases Involving Generators, 2004-2014

Reason

Cases

Weighted

proportion

(%)

Victim realized something was wrong and arranged to get to ER

Discovered in distress by family, friend, or due to a welfare check

Carbon monoxide alarm sounded, arranged to get to ER

Unknown why or how taken to Emergency Room

199

Total

292

100

Table 3 presents a summary of the location of the generator involved with the CO poisoning event. The three most common locations identified were “Inside the home” (33%); “Inside the garage” (25%); and “In the basement” (18%). In 11 percent of the reported cases, the generator was located outside. In half of the “Outside the home” scenarios, the narrative specifically states the location was near a window, door, or air conditioner.

Table 3—Location of Generator in Cases Reported in NEISS CO Poisoning or CO Exposure Cases Associated With Generators, 2004-2014

Generator location

Cases

Weighted

proportion *

(%)

Inside the home

Inside the garage

In the basement

Outside the home

Other/Unknown

Total

292

100

* Percentages do not sum to 100% due to rounding.

The high number of estimated injuries relative to fatalities suggests that many more people leave the scene of the generator, are rescued, or seek care than fatally succumb to CO poisoning. As detailed in subsequent sections, reduced CO emissions will greatly extend the time it takes for CO exposures to result in incapacitation and subsequent death. Moreover, in some cases, reduced CO emissions will actually prevent incapacitation and death from happening, even if an individual does not leave the exposure location. In situations where a generator is operated indoors, the extended window of time will allow exposed individuals a much greater chance of terminating their CO exposure or increase the chance of being found by others before serious injury and/or death can occur. Exposure termination could occur for several reasons, including the following:

• Exposed individuals might leave the exposure location to engage in everyday activities (

e.g.,

work, school),

without necessarily being aware of any developing CO hazard.

• In some cases, exposure termination might occur without the individual leaving the location, simply because the generator runs out of fuel, or power is restored and the generator is shut down in response, which allows CO levels to decay naturally without reaching lethal exposure.

• Exposed individuals might respond to a CO alarm activation.

• Exposed individuals might recognize a growing health concern and leave to seek treatment or summon help (call a friend, relative, or 9-1-1), even if they do not necessarily recognize CO emissions as the cause of early nonspecific adverse health effects of CO poisoning.

• Exposed individuals might be found in an impaired state by other, lesser affected, co-exposed individuals who had been in locations farther away from the generator.

• Exposed individuals might be found by concerned outside parties conducting welfare checks, or by outside parties simply arriving at their home for other reasons, such as, to co-commute to work, a social or official visit, or the return home of a co-occupant from work or school.

The Commission notes that all the reasons specified above for exposure termination have been reported in incidents where there are survivors of carbureted, generator-related CO poisoning. More such cases would be expected with reduced CO emissions, due to an overall downward shift in expected CO poisoning severity. The Commission recognizes that consumers cannot be relied upon to react appropriately to any indication of a CO exposure, and that even those who recognize a developing CO hazard, might decide to enter the area where a generator is located in an attempt to switch it off. This behavior is known to have resulted in lethal outcomes with carbureted generators because CO can accumulate to levels that can cause near-immediate loss of consciousness due to hypoxia/anoxia. However, with reduced CO emissions, the peak CO levels attained in an unventilated area where the generator is operated will be considerably lower than the level that would cause near-immediate loss of consciousness. This potentially could reduce the incidence of death among individuals who enter an unventilated area to turn off a generator, by allowing them time to egress the area before being overcome.

C. Hazard Characteristics

As stated in the previous section, as of May 2015, there were 562 incidents involving fatalities from portable generators reported to CPSC, which occurred between 2004 through 2014. CPSC assigned In-Depth Investigations (IDI) for 535 of these 562 incidents (95 percent), to gather more detailed information about the incident and the product(s) in use. CPSC categorized the incident data in the IDI reports according to the location where the incident occurred:

• 75 percent of deaths (565 deaths, 422 incidents) occurred in a fixed-structure home location, which includes detached and attached houses, apartments, fixed mobile homes, and cabins used as a permanent residence;

• 16 percent (117 deaths, 81 incidents) occurred at non-fixed-home locations or temporary structures, such as trailers, horse trailers, recreational vehicles (RV), cabins (used as a temporary shelter), tents, campers, and boats, and vehicles in which the consumer brought the generator on board or into the vehicle;

• 6 percent (48 deaths, 46 incidents) occurred in external structures at home locations, such as sheds and detached garages;

• 3 percent (21 deaths, 13 incidents) occurred at unknown or other locations.

In the same 11-year period, 42 deaths from 30 incidents

occurred with the generator operating outdoors, where the exhaust infiltrated into a nearby fixed-structure home, a non-fixed-structure home, or temporary shelter.

See Figure 2.

These figures exclude two deaths in 2011 caused by a stationary generator operated outdoors.

Hnatov, Matthew,

Carbon Monoxide Deaths Associated with Engine-Driven Generators Located Outdoors in 2004 through 2014,

U.S. Consumer Product Safety Commission, Bethesda, MD, November 2015. (Docket Identification CPSC-2006-0057-0028, available online at

www.regulations.gov

EP21NO16.004

Of the 565 deaths (422 incidents) that occurred at a fixed structure home:

• 45 percent (256 deaths, 191 incidents) occurred when the generator was operated in the living space

of the house;

Used here, living space includes all rooms, closets, doorways and unidentified areas inside a home, but does not include basements, which are treated as a separate category.

• 25 percent (140 deaths, 108 incidents) occurred when the generator was in the attached garage or enclosed carport;

• 25 percent (139 deaths, 98 incidents) occurred when the generator was in the basement or crawlspace;

• 3 percent (16 deaths, 12 incidents) occurred when the generator was operated outside;

Another 28 deaths from 19 incidents occurred with generators operating outside structures other than fixed-structure home sites, such as RV, camper or trailer, vehicle, boat, or cabin used other than as a permanent residence.

• 2 percent occurred when the generator was at the fixed-structure home site, but exact location was unknown.

See Figure 3.

EP21NO16.005

The reason the generator was needed was identified in more than 80 percent of the 562 incidents. Following are the three biggest causes:

• 27 percent (152 incidents) were associated with the use of generators during a temporary power outage stemming from a weather problem or a problem with power distribution;

• 21 percent of the fatal incidents (116 incidents) were associated with the use of generators after a power shutoff by the utility company for nonpayment of a bill, a bill dispute, or other reason.

• 19 percent of the fatal incidents (109 incidents) did not indicate why the generator was in use, or why there was no electricity at the location of the incident.

Of the 152 fatal incidents associated with a power outage due to weather or a problem with power distribution, 93 percent were due to specific weather conditions. Ice or snow storms are associated with the largest percentage of weather-related CO fatal incidents, accounting for nearly half (49%) of the power outage-related incidents. Hurricanes and tropical storms were associated with 28 percent of CO fatal incidents. More than half (31 of 61) of the generator-related CO fatalities that were hurricane- or tropical storm-related (20 of 42 fatal incidents) occurred in 2005, a year of above-average hurricane activity.

The size of the generator involved in a CO fatality was identified in 45 percent of the 562 incidents. Because most of the generators that were associated with fatal CO poisoning were gasoline-fueled,

staff categorized the size of the generator by using the U.S. Environmental Protection Agency's (EPA) classification of the small SI engine powering it: A handheld engine

; a non-handheld, Class I engine; or a non-handheld, Class II engine.

The incidents involving generators powered by non-handheld, Class II engines were then divided by whether the engine had a single cylinder or twin cylinders.

In the majority of cases (55%), CPSC staff was unable to obtain sufficient information to be able to categorize the generator into one of these classifications. In the incidents where engine classification could be determined, slightly more than one-third (35 percent) involved Class I engine powered generators, and slightly less than two-thirds (63 percent) involved single-cylinder, Class II

engine-powered generators. See Figure 4. There were two incidents involving generators powered by handheld engines that caused one death in each incident. There were three incidents involving generators powered by twin-cylinder, Class II engines that caused seven deaths. Two of the incidents were single-death incidents, and the third incident, with the generator operating outside an RV, caused five deaths inside the RV.

In 52 of the 562 incidents, the fuel type could not be ascertained. Of the 510 cases where the fuel type used in the generator was known, 99 percent (506 of 510) were gasoline-fueled generators. Of the remaining incidents, three involved propane-fueled generators, and the other incident involved a diesel-fueled generator.

Although handheld engines generally are used in equipment that is held or supported by an operator during use (such as trimmers), handheld engines may also be used to power non-handheld equipment, such as smaller portable generators.

The EPA broadly categorizes small SI engines as either non-handheld or handheld, and within each of those categories, further distinguishes them into different classes, which are based upon engine displacement. Non-handheld engines are divided into Class I and Class II, with Class I engines having displacement above 80 cc up to 225 cc, and Class II engines having displacement at or above 225 cc, but with maximum power of 19 kilowatts (kW). Handheld engines, which are divided into Classes III, IV, and V, are all at or below 80 cc.

When the IDI did not report the generator's engine displacement, or it was not obtainable from other information in the IDI, staff considered the power rating of the generator, if the IDI contained information regarding the power rating of the generator. Staff classified generators with a reported wattage of 3.5 kW and larger as powered by a Class II engine and those less than 3.5 kW as powered by either a handheld or a Class I engine. To distinguish the handheld powered generators from the Class I powered generators when there was no information to ascertain the engine displacement, generators with wattage of 2 kW to 3.5 kW were considered to have a Class I engine. To distinguish the single-cylinder Class II engines from the twin-cylinder Class II engines, staff determined from a search of EPA's exhaust emission certification database (

www3.epa.gov/otaq/certdata.htm#smallsi

) that twin-cylinder, class II engines generally have a maximum engine power of nominally 12 kW and higher. Based on manufacturers' generator specifications available online, generators with engines with power equal to or greater 12 kW, typically have a rated power of 9kW and higher. Therefore, staff considered generators with rated power of 3.5 kW up to 9 kW to be powered by a single-cylinder, Class II engine, and those 9 kW and greater to be powered by a twin-cylinder, Class II engine when there was no information to ascertain the engine displacement and number of cylinders.

EP21NO16.006

V. Overview of Proposed Requirements

The proposed standard would apply to portable generators powered by small handheld and non-handheld SI engines. The Commission categorized the size of the generator using the EPA's classification of the small SI engine powering it: A handheld engine, a non-handheld Class I engine, or a non-handheld Class II engine. The Commission further categorized the generators powered by non-handheld Class II engines by whether the engine had a single cylinder or twin cylinders. The Commission defines the

generator

categories (as distinguished from the

engine

categories) as follows:

• A

handheld generator

is a generator powered by an SI engine with displacement of 80 cc or less;

• A

class 1 generator

is a generator powered by an SI engine with displacement greater than 80 cc but less than 225 cc;

• A

class 2 single cylinder generator

is a generator powered by an SI engine with one cylinder having displacement of 225 cc or greater, up to a maximum engine power of 25 kW; and

• A

class 2 twin cylinder generator

is a generator powered by an SI engine with two cylinders having a total displacement of 225 cc or greater, up to a maximum engine power of 25 kW.

Although the Commission categorized generators based on the EPA classification of the engines powering them, it is important to distinguish these engines from the portable generators that they are used in because the engines are also used in other products. To provide a clear distinction, the Commission refers to

engines

according to EPA's classification: Handheld engines, non-handheld Class I engines, and non-handheld Class II engines, while referring to

portable generators

according to the Commission's definitions, handheld generators, class 1 generators, class 2 single-cylinder generators and class 2 twin-cylinder generators.

Generators within the scope of the proposed rule provide receptacle outlets for AC output circuits and are intended to be moved, although not necessarily with wheels. Products that would not be covered by the proposed rule include permanently installed stationary generators, 50 hertz generators, marine generators, generators permanently installed in recreational vehicles, generators intended to be pulled by vehicles, generators intended to be mounted in truck beds, and generators that are part of welding machines. Generators powered by compression-ignition (CI) engines (engines fueled by diesel) are also excluded from the scope of the proposed rule. These inclusions and exclusions are largely consistent with the scope of the two U.S. voluntary standards for portable generators, UL 2201—

Safety Standard for Portable Generator Assemblies

and PGMA G300—

Safety and Performance of Portable Generators.

The great majority of the units that fall within the scope of the proposed standard are gasoline-fueled, but portable generators powered by engines fueled by liquid propane (LP) present similar risks of CO poisoning, and these units also would be covered by the proposed rule. Some portable generators can operate fueled by gasoline, LP and natural gas, and these would also be covered by the scope of the proposed rule.

The proposed rule specifies different limits on weighted carbon monoxide emission rates for different classes of generators in recognition of the effects

of factors such as engine size and other engine characteristics on CO emissions, generator size, weight, and hazard patterns and the different challenges that may be faced in meeting CO emission rates expressed in grams per hour. The performance requirements for the different classes of generators also have a scaling factor of 1.5 applied to the technically feasible rates to account for production variation. Specifically, the proposed rule would require that handheld generators and class 1 generators not exceed a weighted CO rate of 75 grams per hour (g/hr); class 2 single-cylinder generators not exceed a weighted CO emission rate of 150 g/hr; and class 2 twin-cylinder generators not exceed a weighted CO emission rate of 300 g/h. The weighted emission rates are based on weighting of six modes of generator operation, ranging from maximum generator load capability (mode 1) to no load (mode 6), similar to a procedure used by EPA to certify compliance with its emission standards for small SI engines. More detail about this procedure can be found in CPSC's staff briefing package. The performance requirements apply when generators operate at normal oxygen content; however, the Commission remains interested in CO emissions when generators operate at reduced oxygen content of 17 percent. The Commission welcomes comments on the advantages and disadvantages of setting performance requirements at 17 percent oxygen instead of normal oxygen as well as comments on the technically feasible CO emission rates for generators operating at 17 percent oxygen, for each of the generator categories. Furthermore, the Commission welcomes comments on the test methods for CO emissions in both normal oxygen and 17 percent oxygen in Tab J, Appendices A2 and A3 of the staff's briefing package.

The proposed rule does not dictate how generators would meet the CO emission limits. Rather, under the proposed rule, firms have the flexibility to determine the appropriate technology to meet the specified performance requirements. To determine feasibility and to estimate likely costs of the proposed rule, staff's briefing package, and this preamble, discuss ways that staff believes companies might modify generators to meet the CO emission limits. However, companies could use other approaches.

The proposed rule describes the test procedure and equipment that the Commission would use to assess compliance with the standard. Manufacturers, however, need not use this particular test, so long as the test they use effectively assesses compliance with the standard. The Commission believes this approach provides added flexibility to manufacturers to reduce testing burdens. The Commission welcomes comments on the benefits and costs of this approach versus requiring a specifc test method for manufacturers to demonstrate compliance.

In accordance with Section 9 of CPSA, the proposed rule contains a provision that prohibits a manufacturer from “stockpiling,” or substantially increasing the manufacture or importation of noncomplying generators between the date that the proposed rule may be promulgated as a final rule, and the final rule's effective date. The rule would prohibit the manufacture or importation of noncomplying portable generators by engine class in any period of 12 consecutive months between the date of promulgation of the final rule and the effective date, at a rate that is greater than 125% of the rate at which they manufactured or imported portable generators with engines of the same class during the base period for the manufacturer. The base period is any period of 365 consecutive days, chosen by the manufacturer or importer, in the 5-year period immediately preceding promulgation of the rule.

Generator sales can vary substantially from year to year, depending upon factors such as widespread power outages caused by hurricanes and winter storms. Annual unit shipment and import data obtained by CPSC staff show that it has not been uncommon for shipments to have varied by 40 percent or more from year to year at least once in recent years. The anti-stockpiling provision is intended to allow manufacturers and importers sufficient flexibility to meet normal changes in demand that may occur in the period between promulgation of a rule and its effective date, while limiting their ability to stockpile noncomplying generators for sale after the effective date. The Commission seeks comments on the proposed product manufacture or import limits and the base period for the stockpiling provision.

VI. CPSC Technical Analysis and Basis for Proposed Requirements

A. CPSC's Two-Part Prototype Low CO Emission Generator Technology Demonstration Program

CPSC staff developed a two-part technology demonstration program to demonstrate that the small SI engine powering a commercially available portable generator could be modified with existing emission control technology to reduce its CO emission rate to levels expected to reduce the risk of fatal and severe CO poisoning. The objective of the first part of the program was to develop, from a current carbureted engine-driven generator, a prototype with a CO emission rate reduced to the lowest technically feasible level: (1) Without negatively impacting the engine's power output, durability, maintainability, fuel economy, and risk of fire and burn; and (2) while also ensuring that the engine continued to meet EPA's small SI engine exhaust emission standard for hydrocarbons and oxides of nitrogen (HC+NO

), to which the unmodified OEM version of the engine was originally labeled as being certified. For this, CPSC staff sought a target CO emission rate reduction of 90 percent. The objective of the second part of the program was to assess the efficacy of the prototype generator in reducing occupant exposure profiles created by its operation in a fatal scenario commonly reported in CPSC's incident data compared to the exposure profiles created by the unmodified carbureted generator.

Complete documentation on the prototype generator and both parts of the demonstration program is provided in Buyer, Janet,

Technology Demonstration of a Prototype Low Carbon Monoxide Emission Portable Generator,

September 2012. (available online at:

http://www.cpsc.gov/PageFiles/129846/portgen.pdf

and in

www.regulations.gov

in docket identification CPSC-2006-0057-0002).

Part One: Prototype Development and Durability Testing at University of Alabama

The Commission contracted with the University of Alabama (UA) to conduct the prototype development and durability phase of the program. The prototype development started with a commercially available generator with an advertised continuous electrical power output rating of 5.0 kW that was powered by a small, air-cooled, single-cylinder non-handheld Class II carbureted engine with a 389 cubic centimeter (cc) displacement and overhead valve (OHV) configuration. The prototype was a modification of that engine. To develop the prototype, UA replaced the engine's carburetor with a closed-loop electronic fuel-injection (EFI) system, used an oxygen sensor in the exhaust for closed-loop fuel-control feedback, tuned the fuel control to stoichiometry

and replaced the muffler with a muffler that had a small three-way catalyst (TWC) integrated into it. UA subjected the

prototype generator to a durability program for a total of 500 hours, which was the manufacturer's rated useful life of the engine at the time of the program. Simultaneous to the durability program on the prototype generator, UA subjected a baseline unmodified carbureted generator, the identical model to the prototype generator before modification, to the same durability program. UA made periodic emission measurements on both the prototype and the unmodified carbureted generator during the 500 hours of operation to compare the performance of the prototype to the baseline unmodified carbureted generator. After the 500-hour durability program concluded on both the baseline carbureted generator and the prototype generator, an independent laboratory, Intertek Carnot Emission Services (CES), conducted end-of-life emission testing, both with the engine installed in the generator as well as on a dynamometer,

in accordance with the EPA small SI engine test procedures. The purpose of this testing was to ascertain whether, at the end of the engine's rated useful life, the prototype engine's emissions would meet: (1) The EPA's Phase 2 requirements for HC+NO

, and (2) CPSC staff's target reduction for the exhaust CO emission rate.

Stoichiometry is the theoretical air-fuel ratio (AFR) for complete combustion and is the theoretical point for nearly the lowest amount of CO production. AFR associated with stoichiometry for typical gasoline formulations is nominally 14.6.

A dynamometer is an instrument that measures the power output of an engine.

CES's testing in accordance with EPA test procedures showed that the prototype engine, while mounted on a dynamometer and equipped with the muffler that had a catalyst installed, had a 6.0 g/kW-hr CO emission rate. This CO emission rate is 99 percent below the EPA's Phase 2 and Phase 3 CO standard of 610 g/kW-hr.

The prototype engine had an HC+NO

exhaust emission rate of 6.7 g/kW-hr. This rate is 45 percent below the EPA's Phase 2 HC+NO

standard for a Class II engine, to which the engine was originally certified, and 16 percent below the Phase 3 HC+NO

standard that came into effect shortly after CPSC's development program with UA began. CES's dynamometer testing also showed that the prototype engine delivered a maximum power of 7.9 kW, which is within 0.3 kW of the advertised rated power for the unmodified OEM carbureted engine. CES's emission testing of the prototype generator (with the engine still installed in the generator, as opposed to mounted on the dynamometer) measured a weighted CO emission rate of 26.10 g/hr.

Thus, at the end of the engine's rated useful life, the prototype engine's emissions met both EPA's Phase 2 requirements for HC+NO

and CPSC staff's target reduction for the exhaust CO emission rate. Staff's prototype findings have since been repeated by others who patterned their reduced CO emissions prototype generators on the design concept developed for CPSC by the University of Alabama.

Moreover, new generator products with reduced CO emissions, achieved by similar engine design modifications and use of catalysts, are beginning to enter the retail market.

The EPA sets emission standards for all small SI engines. These engines provide power for a wide range of products typically owned by consumers, including portable generators. The EPA's primary emphasis is on regulating emissions that contribute significantly to nonattainment of the National Ambient Air Quality Standards (NAAQS) for ozone, of which hydrocarbons and oxides of nitrogen (HC+NO

) are precursors. For non-handheld engines, the EPA adopted emission standards referred to as Phase1 in 1995, Phase 2 in 1999, and Phase 3 in 2008.

The highest of three tests was 26.10 g/hr. The other two tests yielded weighted CO rates of 23.47 and 19.38 g/hr.

See Techtronic Industries (TTi) presentation on 3/17/16, at PGMA's Technical Summit on Carbon Monoxide Hazard Mitigation for Portable Generators—pages 85-105 of 178 page pdf file at:

http://www.cpsc.gov//Global/Newsroom/FOIA/Meeting%20Logs/2016/MeetingLogPGMA31716.pdf.

See Tab I staff's briefing package.

Part Two: Comparative Testing of Unmodified Carbureted (Baseline) and Prototype Generators at National Institute for Standards and Technology

The Commission entered into an interagency agreement with NIST to conduct the second part of the program. In this part of the demonstration program, NIST operated one generator in its unmodified carbureted configuration and another generator in the prototype configuration in the attached garage of a test house on NIST's campus. The test house is used for conducting indoor air quality (IAQ) studies. NIST measured the CO accumulation in the garage and transport into the house. The results provide a sense of how quickly a commonly fatal consumer scenario develops with an existing carbureted generator, and what the comparative results are from the same tests with the fuel-injected catalyzed prototype.

Another objective of the IAG was to determine each generator's mass CO emission rates at each of the six loads used in the load profile. This work also supported NIST's validation of NIST's multizone airflow and contaminant transport model CONTAM, which is used to predict contaminant concentrations throughout a modeled structure resulting from a source mass emission rate located somewhere within the structure. NIST used CONTAM in predicting the health effects of the CO rates associated with the proposed performance requirements.

NIST compared the garage CO concentrations from the prototype and the unmodified carbureted generator, after equal periods of generator run-time in the tests, with the garage bay door fully closed. NIST found that the prototype showed 97 percent reduction in the amount of CO released into the garage, compared to the unmodified carbureted generator. This reduction is consistent with UA's findings and translated to much lower levels of CO transporting throughout the house. Taking into consideration the CO time course profile (which is the CO concentration over time) of each room of the house and of the garage, the Commission performed health effects modeling and estimated that the prototype generator resulted in a significantly extended time interval for hypothetical occupants to escape or to be rescued before being incapacitated. For example, in one test in which the garage bay door and connecting door to the house were both closed, the time interval increased by a factor of 12 with the prototype, compared to the unmodified carbureted generator (from 8 minutes to 96 minutes) for the deadly scenario of a consumer in the garage with the generator. The time interval increased even more for occupants inside the house.

The Commission believes that this increased time interval could give occupants an opportunity to remove themselves from the exposure before being incapacitated (perhaps due to their symptoms or other reasons such as an unrelated need to leave the house) or to be found alive by others. In contrast, the Commission predicts that the high CO emission rate of the unmodified carbureted generator would cause some of the occupants, depending on where they are located, to experience relatively quick onset of confusion, loss of muscular coordination, loss of consciousness, and death, without having first experienced milder CO poisoning symptoms associated with low or slowly rising CO-induced hypoxia.

B. Staff Assessment of Feasible CO Rates Based Upon EPA's Technology Demonstration Program and Staff Testing of Fuel-Injected Generators

A technology demonstration conducted by EPA further demonstrates the feasibility of significantly lowering CO emission generators using EFI.

2006, EPA examined the feasibility of reducing HC+NO

emissions beyond their Phase 2 standards.

EPA applied EFI and high-efficiency catalysts on two single-cylinder, air-cooled engines, both nominally 500 cubic centimeters (cc) in displacement with overhead valve (OHV) configurations. Because CO and NO

emissions have an inverse relationship, in focusing on reducing HC+NO

emissions, EPA specifically chose to test with catalysts formulations designed to

minimize

CO oxidation.

McDonald, Joseph, Olson B, and Murawski M,

Demonstration of Advanced Emission Controls for Nonroad SI Class II Engines,

SAE paper 2009-01-1899; McDonald, Joseph, Memorandum, Re: Supplemental Engine Dynamometer Data, May 5, 2006. (available online in:

www.regulations.gov

docket identification EPA-HQ-OAR-2004-0008-0372.).

U.S. EPA,

Control of Emissions from Marine SI and Small SI Engines, Vessels, and Equipment—Final Regulatory Impact Analysis,

EPA420-R-08-014, September 2008 (available online in

www.regulations.gov

in docket identification EPA-HQ-OAR-2004-0008-0929); U.S. EPA,

EPA Technical Study on the Safety of Emission Controls for Nonroad Spark-Ignition Engines <50 Horsepower,

EPA420-R-06-006, March 2006, Docket Identification EPA-HQ-OAR-2004-0008-0333. (available online at: (

http://www.epa.gov/nonroad/equip-ld/phase3/420r06006-rpt-2appdx.pdf

Oxidation of CO to carbon dioxide (CO

) is the means by which CO emissions are reduced in a catalyst.

EPA used low-cost engine management and fuel injection systems that were similar to that which UA used for the CPSC prototype generator. While the UA generator prototype used a closed-loop system and tuned the fuel to stoichiometry at the high loads, in interest of cost-savings, the EPA engines did not use an oxygen sensor necessary to make it a closed-loop fuel system. For its engines, EPA replaced the carburetor with open-loop EFI that was calibrated rich of stoichiometry,

i.e.,

a lower air-to-fuel ratio, at moderate-to-high loads and near stoichiometry at light load conditions to achieve the desired emission control of HC+NO

. EPA developed integrated catalyst-muffler systems for its engines, all selected to prioritize NO

reduction and HC oxidation over CO oxidation. Even though EPA was intentionally trying to select catalysts that would

minimize

CO oxidation, both engines achieved an average 68 percent reduction in the weighted CO emission rate. The average of the weighted CO emission rate of the two carbureted OEM configurations was 1,760 g/hr, and the average of the two EFI configurations with the catalyst providing the most reduction in CO emissions was 565 g/hr.

Although the EPA noted that some engines may need improvements to accommodate stoichiometric fuel control (such as redesign of cooling fins, fan design, combustion chamber design, and a pressurized oil lube system), EPA concluded that closed-loop EFI with fuel control at or near stoichiometry

technically feasible and

is not

cost prohibitive on all Class II engines.

U.S. EPA,

EPA Technical Study on the Safety of Emission Controls for Nonroad Spark-Ignition Engines <50 Horsepower,

EPA420-R-06-006, March 2006, Docket Identification EPA-HQ-OAR-2004-0008-0333. (available online at: (

http://www.epa.gov/nonroad/equip-ld/phase3/420r06006-rpt-2appdx.pdf

CPSC staff believes that with a focus on reducing CO emissions, a lower weighted CO emission rate could have been achieved by using an oxygen sensor for closed-loop feedback, operation closer to stoichiometric at the higher loads, and a different catalyst formulated for higher conversion efficiency of CO.

See CPSC staff's briefing memorandum and Tab I of the briefing package for a more detailed explanation.

CPSC staff tested three fuel-injected generators created by three different manufacturers.

Two of these generators, neither of which was designed for low CO emissions, are available in the marketplace, and the third is a manufacturer's prototype generator that was designed for low CO emissions. The first of the three generators is a 10.5 kW rated generator powered by a twin-cylinder Class II engine with nominal 700 cc displacement and overhead valve (OHV) configuration. The generator does not have a catalyst for aftertreatment and the generator's engine is calibrated rich of stoichiometry at higher loads and at stoichiometry with closed-loop fuel control at moderate-to-light load conditions. Based on CPSC staff's testing of this generator in normal atmospheric oxygen, which found a 670 g/hr weighted CO emission rate, as well as on staff's engineering assessment of its physical and operational characteristics, staff believes that it is reasonable to expect that this engine could operate closer to stoichiometric at the higher loads and that a catalyst formulated for some CO conversion efficiency could be used for aftertreatment to further reduce its CO emission rate to nominally 200 g/hr.

See Tab I of the staff's briefing package.

The second generator is a 5.5 kW rated power generator powered by a single-cylinder Class II engine with nominal 400 cc displacement and OHV configuration, equipped with an oxygen sensor for some form of partial closed-loop operation and a catalyst. The engine is calibrated rich of stoichiometry at all loads. Based on staff's testing in normal atmospheric oxygen that found a nominal weighted CO rate of 560 g/hr, staff believes a CO emission rate of nominally 100 g/hr is possible, if the generator were operated closer to stoichiometric for at least some of the loads and used a catalyst formulated for higher CO conversion efficiency.

The third generator is a 5.5 kW rated power generator powered by a closed-loop fuel-injected single-cylinder Class II engine with nominal 400 cc displacement and OHV configuration. It has a catalyst for aftertreatment and the engine is calibrated to stoichiometric AFR with closed-loop operation at all loads. Staff's testing of this generator in normal atmospheric oxygen found a weighted CO rate of 81 g/hr.

C. Assessment of Epidemiological Benefits of Reduced CO Emission Portable Generators—NIST CO and COHb Modeling Study

1. Background

To assess the epidemiological benefits of reduced CO emission generators, CPSC contracted NIST to perform a series of CO exposure simulations that would model the operation of a portable generator in various locations within various house configurations and other structures, and at various CO emission rates.

CPSC used these results to determine the possible deaths averted if reduced CO emission generators had been used, as described below.

Emmerich, Steven J., B. Polidoro, W. Dols, Simulation of Residential CO Exposure Due to Portable Generator Operation in Enclosed Spaces (NIST Technical Note 1925), 2016.

2. CO Emission Modeling

NIST modeled 40 different structures, including houses with basements and others with crawlspaces, as well as ones with slab-on-ground construction, with and without attached garages, and including older construction and newer construction homes. Three different external residential structures designed to represent detached garages and sheds were included in the 40 structures. The 37 different house models included detached home, attached home, and manufactured home designs. House models and other structures used in the modeling study were matched to 503 out of the 659 actual generator-related CO fatalities reported to CPSC over the period 2004 to 2012. One hundred fifty-six fatalities (659 minus 503) were not included in the modeling analysis because the generator was either outdoors or in a structure such as a camper, RV, tent, church, boat, or apartment complex that was not similar to any of the structure models used by NIST. The Commission believes that reduced emission generator use in these scenarios would most likely have produced fewer CO fatalities than the number observed in the incident data.

This would be especially true in scenarios with the generator running outdoors, or in a large-volume space, such as a church.

CPSC staff chose the modeled CO emission rates based on: (1) CPSC's estimates of elevated CO emission rates expected for the four categories of current carbureted generator products when operating in a reduced oxygen environment, and (2) a series of reduced CO generation rates that allowed CPSC to assess benefits and costs of various levels of reduced emissions within technically feasible rates for each generator category.

The first part of the modeling study used the NIST multizone airflow and contaminant transport model CONTAM, which predicted CO levels in different areas of each structure, over a 24-hour period.

Determination of CO Emission Rates, Run Times, and Heat-Release Rates for Carbureted Generators

Staff determined CO emission rates, run times, and heat release rates for NIST to model for current, carbureted generators (baseline carbureted generators) based on data from EPA's non-road small spark-ignition engine (NRSI) certification data Web site and advertised power ratings and engine specifications for representative products. These baseline parameters are shown in Table 4, and an explanation of the basis for the parameters follows.

Table 4—Modeled CO Emission Rates, Run Times, and Heat-Release Rates for Baseline Carbureted Generators

Generator category

Average

weighted

CO rate at

17% O

(g/hr)

Average

run time

(hrs)

Average heat

rlease rate

(kW)

Handheld

900

Class 1

1,800

Class 2 Single Cylinder

4,700

Class 2 Twin Cylinder

9,100

To determine values for CO emission rates, run times, and heat-release rates representative of current generators involved in the fatal incidents, staff considered the generators produced by six large generator manufacturers. All of these manufacturers are members of the Portable Generator Manufacturers Association (PGMA), and, as documented on PGMA's Web site, are the major manufacturers of portable generators sold in North America and a significant majority of the industry.”

Staff used the manufacturers' reported product specifications for 31 generators ranging from 900 to 15,000 watts rated power and developed the representative parameters for each of these inputs based on the range of generators in each of the four categories in Table 4.

www.pgmaonline.com.

Staff used the engine specifications provided by the generator manufacturer to search the EPA's NRSI engine certification data Web site to find the published CO emission rate corresponding to each generator's engine. Staff then calculated the weighted CO emission rate (in g/hr) for each generator's engine, by multiplying the g/kW-hr rate by 46.7 percent of the maximum engine power (46.7 percent of the maximum engine power is the weighted average based on the EPA six-mode calculations).

Staff assumes that the typical load profile of a portable generator used by a consumer is that of the weighted profile. In addition, staff assumes the engine's weighted CO rate is that of the generator.

The engine manufacturer's CO emission rate reported in the EPA's exhaust emission certification Web site, in terms of grams per kilowatt-hour (g/kW-hr), is the sum of six weighted CO rates in grams per hour (g/hr) that the engine emits while installed on a dynamometer test platform and operating with each of six steady-state loads applied (also referred to as modes) divided by the sum of the weighted power for those six modes. The EPA's six-mode test cycle was developed with industry to replicate typical in-use operation of small utility engines when used in all types of engine-driven products.

Considering that 95 percent of the generator-related CO fatalities in CPSC's databases occurred when the generator was operated in an enclosed space, it is important for modeling studies to consider the CO emission rate when a carbureted generator is operating in such enclosed space scenarios. Evidence supporting this view is seen in results of findings from generator tests conducted by NIST under a prior interagency agreement with CPSC.

NIST's tests, as well as subsequent staff testing, showed that the CO emission rate of current carbureted generators increases threefold as the oxygen drops from normal levels (approximately 20.9 percent oxygen) to approximately 17 to 18 percent oxygen when a generator is operated in an enclosed space, such as those reported in the incident data. Consequently, to reflect more accurately current carbureted generator operation under oxygen depletion conditions, staff's calculated weighted CO emission rate, when each generator is operated outdoors at normal oxygen, was multiplied by a factor of 3.

Emmerich, S.J., A. Persily, and L. Wang,

Modeling and Measuring the Effects of Portable Gasoline-Powered Generator Exhaust on Indoor Carbon Monoxide Level

(NIST Technical Note 1781), Feb 2013.

The generators' run time on a full tank of gas that was associated with 50 percent of the advertised rated load was used to determine the full-tank run time used in the modeling. Fifty percent load was used because, as stated above, 46.7 percent of the engine's maximum power represents the weighted load profile, which is nominally 50 percent. Staff generally used manufacturer's product specifications for run time at 50 percent load, and in a few cases, used engineering estimates to determine the run times. Staff chose to model run times based on a full tank of fuel as a conservative assumption, despite knowledge of scenarios where a generator was used to allow completion of a specific short-duration task, in temporary power outage situations where power was restored within a few hours before a full tank of fuel could be consumed, or in scenarios where the generator was still running when victims were found, had summoned help, and/or had removed themselves from the area.

Staff estimated heat-release rates for these generators based on the fuel-consumption rate at 50 percent load, the manufacturer's specification for the generator's tank capacity, a heat of combustion of gasoline of 42.5 MJ/kg, and an assumed conservative 35 percent thermal efficiency of the engine.

Determination of CO Emission Rates, Run Times, and Heat Release Rates, for Reduced Emission Rate Portable Generators

NIST used the same values for run times and heat-release rates for the reduced CO emission rates of each generator category as those used for current generators.

NIST modeled the rates of 50, 125, 250, 500, 1,000 and 2,000g/hr. The three lowest of these approximates the range of CO emission rates that staff believes are technically feasible for both the handheld and class 1 generator categories (50 g/hr), class 2 single-cylinder category (100 g/hr), and class 2 twin-cylinder category (200 g/hr) in ambient air with normal atmospheric oxygen.

CPSC staff reasons that an additional weight and volume of the emission control components needed to reduce the CO emission rate, could be offset by a smaller fuel tank and due to the improved fuel efficiency of reduced emission engines, the smaller tank would still be able to maintain similar run times to carbureted units with larger fuel tanks.

Weather, Temperature and CO Rate Parameters for Carbureted and Reduced CO Emission Generators

Simulations were run for each model structure and model generator location for 28 representative weather days to determine the CO time course profiles, which are the minute-by-minute CO concentration levels in each of the various rooms of the house. The 28 weather days were chosen to include 14 cold weather days (Detroit, MI), seven weather days from warm months (Miami, FL) and seven transition months weather days (Columbus, OH) to represent the distribution of fatalities, which has been seen to skew towards cold-weather days in a similar manner.

Starting indoor temperatures were assumed to be 23 °C in all rooms, and temperatures were modeled to change within the rooms, based on heat transfer related to the heat release from the generator. Thus, generators of various sizes were modeled to be running on 28 different weather days for a full-tank run time

in various rooms within each of the structures, with run times and heat-release rates appropriate to that size of generator, and emissions based on current carbureted generators, or based on possible reduced-emission generators for comparison. In the modeling of baseline carbureted generators, to simulate the increasing CO emission rate as the oxygen level drops in the space the generator is operating (and thus, a lower CO emission rate at the beginning of operation than later), NIST modeled CO rates for the first 2 hours of operation that were only two-thirds of the rates shown in Table 4. After 2 hours, the CO rates were increased to the rates in Table 4 for the duration of the run time. In contrast, as another conservative assumption, NIST modeled

reduced

CO emission rates as constant rates for the entire respective generator run time. The results of the models provided CO time-course profiles for each room of each structure on each weather day for each generator type and location and emission rate.

The 28 individual days were selected using historic weather data recorded at three different geographic locations and three different temperature ranges to approximate the distribution of incidents observed in the CPSC incident data at a generalized level. Although the weather days may be consecutive (

e.g.,

14 consecutive cold weather days), there was no carry-over effect from one day to the next. Each day modeled was reset to zero CO. Therefore, each day, from a CO standpoint, was an independent event.

NIST also modeled half-tank run times to simulate scenarios where shorter duration were considered more appropriate (

e.g.,

in scenarios in which the generator was being used to allow completion of a specific short-duration task at an unpowered location, in temporary power outage situations, where power was restored within a few hours before a full tank of fuel could be consumed, or in scenarios where the generator was still running when victims were found, had summoned help, and/or had removed themselves from the area). While staff has these modeling results, staff only analyzed the modeling results for the full-tank run times to estimate those benefits so as to be consistent with a conservative estimate of benefits.

3. Application of COHb Modeling

The second part of the modeling study used the CONTAM-generated CO time course profiles as input values to predict corresponding COHb levels expected in healthy adults, as a function of time, using Coburn Forster Kane (CFK) modeling.

Conservative assumptions were made about respiratory rates, given expected activity rates over the 24-hours of modeled exposure. The respiratory minute volume (RMV), expressed in liters per minute (L/min), is the specific inhalation rate input value used in the CFK, and for the epidemiological benefits calculated in this analysis, staff used an RMV of 10 L/min. Staff's use of a constant 10 L/min RMV for light activity likely overestimates the breathing rate (and CO uptake rate) of a significant number of victims. In the majority of fatal incidents, victims were at home during an unplanned power outage, or an outage due to utility shut off, and there was no indication that they had engaged in more than sedentary-to-light activity levels for most of the time. For example, in several of these cases, a generator was first started in an enclosed space late in the evening/night at a time where victims were clearly preparing for/or retired to bed; in these instances, a sedentary/resting activity level of 6 L/min RMV would be more appropriate. Thus, use of an RMV of 10 L/min is another conservative assumption in the analysis. This is explained in more detail in Tab K of staff's briefing package and its appendix.

The CFK modeling is a nonlinear differential equation that is a physiologically based mechanistic model for predicting CO uptake and COHb formation and elimination in humans; it has been validated by empirical data from human studies and is widely regarded by authoritative sources as a reasonably reliable and broadly applicable COHb model for acute exposures.

To assess the impact of low-emission generators on potential reductions in CO fatalities, the number of observed fatalities from the incident data were assigned to one of the model structures. The initial step was to assign the fatalities that occurred in an “exact match” structure type. “Exact match” structures are defined as those that match all of the NIST structure characteristic parameters used in the analysis to describe the structure, such as floor area, number of floors, existence of a garage and/or basement. Where exact matches could not be assigned, fatalities were apportioned among best matching structure types (those matching the most number of NIST parameters).

These simulations included various generator location scenarios, dependent on house/structure model designs (

i.e.,

only models that had a basement included the generator-in-basement scenario; and only models that had an attached garage included the generator-in-the-attached garage scenario). To match, as closely as possible, actual usage patterns, the simulation results of the generator locations within the house/structure were proportionately equal to those observed in the incident data.

The victim's location in the modeled house is assumed to have equal probability of occurring in any living space room. This assumption was made for three reasons. In multi-fatality incidents, victims were often found in different locations within a house. In many cases, the victim's location could not be determined from available reports. Moreover, it was frequently unclear whether victims were located in a single area in which they were found for the entire time or if the individual moved around through various parts of the structure. An example of the latter case could be that an individual felt sick and moved, perhaps, to a bedroom to lie down before expiring.

Next, CPSC staff incorporated criteria that staff developed to evaluate modeled COHb profiles considered indicative of

fatal versus nonfatal outcomes. CPSC's Health Sciences (HS) staff developed four “COHb Analysis Criteria” to assess whether predicted COHb profiles from modeled residential scenarios were likely indicative of fatal or nonfatal CO exposures in average adults.

Where a fatal outcome is predicted, the criteria can be used to assess the predicted time to reach fatal exposure during a 24 hour modeling period for each simulated CO exposure. The criteria are intended to reflect the fact that lethal CO health effects are not simply a function of acute hypoxia resulting from a critical reduction in blood levels of oxygen delivered to tissues, as indicated by attainment of a specific peak COHb level.

The criteria include some consideration of the level and duration of the predicted COHb elevation, which recognizes that, in addition to reducing oxygen delivery to tissues, CO can enter the non-vascular body compartment and adversely impact important cellular functions by displacing oxygen from various intracellular heme proteins (particularly myoglobin proteins found predominantly in cardiac and skeletal muscles, and certain cytochrome P-450 enzymes involved is cellular respiration). In some prolonged CO elevations, the additional nonvascular adverse effects of CO can result in death at COHb levels that are not typically lethal.

See Tab K and Tab K appendices of staff's briefing memorandum.

Oxygen binding sites of hemoglobin molecules have more than 200-fold higher affinity for CO than for oxygen.

Although the relationship is not absolute, physiological, epidemiological, and clinical studies provide evidence that acute CO poisoning effects in healthy adults tend to follow toxicological dose-response principles, and that risk of more serious adverse CO poisoning effects worsen progressively as blood levels of COHb increase.

However, it is clear that lethal CO exposures cannot be defined simply by attainment of a single COHb level. Staff used several information sources to develop COHb assessment criteria to facilitate calculation of benefits estimates predicted for generators with reduced CO emissions. A recent authoritative review of CO toxicity by the Agency for Toxic Substances and Disease Registry indicates that there is a high risk of lethal outcome once COHb levels have reached a critical window, which, for healthy individuals, is generally considered to lie between 40 percent and 60 percent COHb.

HS staff reviewed information on COHb levels of victims who experienced acute, generator-related CO poisoning; COHb levels documented in fatal CO poisoning cases reported to CPSC were compared with COHb levels reported for a select group of survivors who received hyperbaric oxygen treatment (HBO-T) for generator-related CO poisoning injuries considered to be of high severity. Staff also considered information on fatal and nonfatal COHb levels reported in non-fire-related CO poisoning cases that did not specifically involve generator-related CO exposures. Based on review of available data on COHb levels in fatal and nonfatal generator-related CO exposures, and other non-generator, non-fire related CO deaths and injuries, staff developed the following criteria to distinguish between modeled COHb levels indicative of lethal versus nonlethal outcome:

For example, loss of consciousness is not generally expected in average adults if peak COHb levels remain below 20 percent, but becomes increasingly more likely as levels approach, and exceed, 40 percent COHb. (Note: Staff is referring to the acute COHb blood levels actually reached, or predicted by modeling, which is not necessarily the same as the highest measured COHb levels reported in clinical cases, where initial COHb measurements are typically reduced from peak levels attained, primarily due to the time lag between the end of CO exposure and blood sampling, plus use of supplemental oxygen during this interval).

Agency for Toxic Substances and Disease Registry (ATSDR), (June 2012) Toxicological Profile for carbon monoxide (web link:

http://www.atsdr.cdc.gov/toxprofiles/tp201.pdf.

(1) If peak level is ≥60% COHb, assume death.

(2) If peak level is ≥50% COHb but <60%, assume death unless average duration of elevation >50% COHb is less than 2 hours, and average duration of elevation between ≥40% and <50% COHb is less than 4 hours.

(3) If peak level is ≥40% COHb, but <50% COHb, assume death if duration of the average in this range exceeds 6 hours.

(4) If peak level is ≤40% COHb, assume survival.

4. Determination of Deaths Averted

The final part of the modeling study used patterns evident in fatal incident data (such as the known percentages of deaths related to various generator locations for various generator sizes and structure types) to modulate the modeled COHb data to estimate the number of fatal CO exposures reported for each generator category that could have been averted at each reduced emission rate. The modeling included exposure duration of up to 24 hours, estimated on a minute-by-minute resolution, and determined the status of living versus dead for modeled occupants at each minute in time. The model assumed equal probabilities of intervention over a 24-hour period. This assumption was used because frequently, one could not determine from the incident data how long of an interval between when the generator was started and when the victim died or some other type of intervention occurred.

Although CPSC incident data reflect primarily fatal CO incidents, the assumption that surviving people eventually depart the exposure is supported by staff's estimates of at least 25,400 medically attended CO injuries involving generators over the period of the deaths modeled and the fact that in some fatal incidents, there were surviving victims. For each scenario (CO emission rate, structure model, generator location, occupied zone, weather day), the model produced estimated COHb levels. From these COHb levels, staff determined at each minute interval, whether the victim was dead or alive, based on the criteria outlined above. The average per-minute interval over the 28 days produced a probability of fatality at the given time. Under the assumption of equal probability of intervention over the 24-hour period, the average probability of fatality over the 24-hour period is the overall fatality rate for the given scenario. For the current carbureted generator model simulation, the probability was normalized (scaled up) to 100 percent of the allocated deaths because this is based on the actual incident data. The reduced emission rate simulation results were scaled up by the same factor to normalize the data. The difference between the allocated deaths per scenario and the number estimated for the reduced emission levels is the estimate of the deaths averted for the specified scenario. The summation of all the modeled scenarios (at a given emission level) represents an estimate of the potential deaths averted, if a reduced emission level generator had been in use in place of the current carbureted types. Thus, the same scenarios and assumptions were used for each generator size, generator location, structure, and weather day combination for current and reduced emissions generators so that the comparison was consistent and the assumptions would apply in the same way to current and reduced emissions.

Table 5 presents a summary of the number of deaths that potentially could have been averted over the 2004 to 2012 time span, if low-emission generators were used in place of the high CO output generators that were in use during this period. CPSC staff estimates that a total of 208 out of 503 deaths

could have been averted. CPSC staff realizes there is uncertainty associated with this estimate given the assumptions and estimations staff used in developing this estimate. However, CPSC staff used conservative values and believes the uncertainty in the estimate is within the range of the sensitivity analysis that staff performed on the effectiveness of the emission rates, as described in the preliminary regulatory analysis.

Table 5—Summary of Potential Deaths Averted at Technically Feasible CO Emission Rates in Reduced Oxygen, 2004-2012

Generator category

CO emission

rate * simulating

generator

operation in

an enclosed

space

Actual

fatalities

allocated

by class

Potential

deaths

averted

Potential

lives saved

rate

(%)

Handheld

150

3.7

1.7

46.6

Class 1

150

176.2

87.7

49.7

Class 2 Single Cylinder

300

321.3

117.9

36.7

Class 2 Twin Cylinder

600

1.8

0.3

17.2

Total

503.0

207.6 = ~208

41.3

* These rates are 3 times the technically feasible rates at normal ambient oxygen (~20.9%) to account for CO emission rate increase in reduced oxygen. To account for production variation the CO emission rates in the proposed requirements are 1.5 times the technically feasible rate in normal oxygen.

The numbers are based on the conservative assumption of CO emission rates tripling from technically feasible rates in normal oxygen for each generator category when operating in theorized oxygen depletion. Staff tripled the rates because staff determined that in reduced oxygen levels, the emission rates of generators that meet the technically feasible rates in ambient air may increase. This factor of 3 is based on testing of carbureted generators conducted by NIST

and CPSC staff.

However, test results from NIST

indicate that the EFI generator depleted the oxygen significantly less than the carbureted generator when tested in each matched pair identical test scenario. Furthermore, based on staff's testing of three generators with fuel-injected engines having different degrees of closed-loop operation, staff believes the factor of increase when the oxygen is 17 percent may be less than 3 for some generators that use closed-loop EFI.

Therefore, based on both of these issues, the factor of 3 could likely overstate the weighted CO emission rates for some EFI generators when operated indoors, and understate the reduction in deaths and injuries resulting from the proposed standard. Consequently, staff believes that the assumption of a threefold increase in the technically feasible rates in ambient oxygen is an appropriate assumption to model, conservatively, for generators operating in enclosed space. Thus, staff ultimately determined epidemiological benefits overall, based on emission rates of 150, 300, and 600 g/hr technically feasible rates, as shown in Table 5.

Emmerich SJ, Polidoro, B, Dols WS.

Simulation of Residential CO Exposure Due to Indoor Portable Generator Operation,

NIST Technical Note 1925, 2016.

See Tab J in the staff's briefing package.

Buyer J.

Technology demonstration of a Prototype Low Carbon Monoxide Emission Portable Generator.

U.S. Consumer Product Safety Commission, Bethesda, MD, September 2012.

Tab J of staff's briefing package.

Staff expects that some additional, but unquantified deaths, could be averted in the remaining 24 percent of fatalities that were not modeled, especially in fatal incidents where a generator was operated outdoors, and/or, that had co-exposed survivors. Staff's epidemiological benefits analysis is contained in TAB K of the staff's briefing package.

VII. Relevant Existing Standards

A. Portable Generator Label

On January 4, 2007, the CPSC voted unanimously (2-0) to require manufacturers of portable generators to warn consumers of carbon monoxide (CO) hazards through a mandatory label containing performance and technical data related to the performance and safety of portable generators. The required warning label informs purchasers: “Using a generator indoors CAN KILL YOU IN MINUTES”; “Generator exhaust contains carbon monoxide. This is a poison you cannot see or smell”; “NEVER use inside a home or garage, EVEN IF doors and windows are open”; “Only use OUTSIDE and far away from windows, doors, and vents.” The label also includes pictograms. The label requirement went into effect on May 14, 2007, and is required for any portable generator manufactured or imported after that date.

Although the Commission believes that the mandatory label for portable generators might prevent some incidents of CO poisoning and death, as discussed in more detail in Section VIII of this preamble, evidence suggests that labeling alone is not sufficient to address the CO poisoning hazard, and that performance requirements for portable generators are needed.

16 CFR part 1407.

B. Voluntary Standards

Underwriters' Laboratories Inc. (UL) and the PGMA have each been accredited by the American National Standards Institute (ANSI) to develop a U.S. safety standard for portable generators. However, only PGMA has developed an ANSI standard for portable generators, ANSI/PMGA G300-2015. UL has also developed a standard, UL 2201, which has not become an ANSI standard, due to lack of consensus. International Organization for Standardization (ISO) 8528-13:2016,

Reciprocating Internal Combustion Engine Driven Alternating Current Generating Sets—Part 13: Safety,

is a standard applicable to portable generators sold overseas.

1. UL 2201

In 2002, UL formed a standards technical panel (STP) to develop the first voluntary standard in the United States, dedicated solely to portable generators, UL 2201

Safety Standard for Portable Generator Assemblies.

CPSC technical staff joined the STP for UL 2201 at its inception and has been an active participant with a long record of

advocating that the standard address CO poisonings.

The requirements in UL 2201 cover internal combustion engine-driven generators rated 15 kW or less, 250 V or less, which are provided only with receptacle outlets for the AC output circuits. The scope section of UL 2201 states that the standard addresses: “the electric shock, fire, and casualty aspects associated with the mechanical performance and the electrical features of portable engine-driven generator assemblies.” The standard restates the mandatory CPSC label requirement, but the standard does not otherwise address the risks related to CO poisoning. UL 2201 includes construction requirements to define minimum acceptability of components of the fuel system, engine, alternator, output wiring and devices, frame/enclosures and others, to ensure their suitability in this application to mitigate the risk of shock, fire and physical injury to users. The standard includes tests applicable to electrical, fire or mechanical hazards, as well as manufacturing tests.

UL has been unable to achieve consensus within the STP for UL 2201 to be recognized as an ANSI standard. Therefore UL 2201, first published in 2009, currently exists as a UL standard without ANSI recognition.

In January 2014, CPSC staff sent a letter to the UL 2201 STP Chair to request that a task group be formed to work on proposals to address the CO hazard that would eventually be balloted by the STP.

The letter outlined a framework of requirements based on work done by and for CPSC staff, which could be used as a starting point for discussions. This letter is described in more detail in the staff's briefing package. Accordingly, UL formed a task group with a roster of 37 members representing a broad range of stakeholder interests, including manufacturers of engines, generators, fuel-control systems and emission control components; public health officials; first responders; medical experts; indoor air quality experts; and government representatives from National Institute for Occupational Safety and Health (NIOSH), Centers for Disease Control and Prevention (CDC), NIST, and CPSC staff. The task group chair is a representative from NIOSH. The first meeting of the task group was held in May 2014. As of August 2016, there have been 26 meetings, all held as teleconference meetings, and there has been active participation and constructive input from a number of the members, but the task group has not yet sent a proposal to the STP to consider for adoption into UL 2201. A more detailed description of this effort is provided in TAB I of the staff's briefing package.

Buyer, Janet, letter to Diana Pappas-Jordan, RE: CPSC Staff Request for Formation of a Working Group and Staff's Recommendations for Requirements to Address the Carbon Monoxide Poisoning Hazard Associated with Portable Generators, January 14, 2014.

http://www.cpsc.gov/Global/Regulations-Laws-and-Standards/Voluntary-Standards/Portable-Generators/CPSCstafflettertoULdatedJan142014.pdf.

The Commission is unaware of any portable generator that is, or has been, certified to UL 2201; as such, it is unlikely that there would be substantial compliance with the standard it if CO emissions requirements were incorporated.

2. ANSI/PGMA G300-2015

In 2011, PGMA was accredited by ANSI to be a standards development organization, allowing PGMA, in addition to UL, to develop a standard for portable generators. PGMA is the accredited standards development organization for ANSI PGMA G300—

Safety and Performance of Portable Generators.

CPSC staff served on PGMA's canvass committee. CPSC staff submitted comments to the standard, including comments regarding the lack of requirements in the standard to address the CO hazard.

PGMA published the first edition PGMA G300 as an American National Standard in June 2015.

Buyer, Janet, letter to Joseph Harding, Subj: CPSC Staff Comments on BSR/PGMA G300-201x,

Safety and Performance of Portable Generators,

January 2, 2015.

http://www.cpsc.gov/Global/Regulations-Laws-and-Standards/Voluntary-Standards/Portable-Generators/CPSCstafflettertoPGMAregardingG300draftstandarddated122015.pdf;

Buyer, Janet, letter to Joseph Harding, Subj: CPSC Staff Comments on BSR/PGMA G300-201x,

Safety and Performance of Portable Generators

dated January 30, 2015, March 6, 2015.

http://www.cpsc.gov/Global/Regulations-Laws-and-Standards/Voluntary-Standards/Portable-Generators/CPSC-staff-letter-to-PGMA-with-comments-on-draft-G300-standard.pdf.

PGMA G300 provides a method for testing the safety and performance of portable generators “rated 15 kW or smaller; single phase; 300 V or lower; 60 hertz; gasoline, liquefied petroleum gas (LPG) and diesel engine driven portable generators intended for multiple use and intended to be moved, though not necessarily with wheels.” PGMA G300 includes construction requirements for engines, fuel systems, frame/enclosures, alternators, and output wiring and devices. The standard includes safety tests intended to address electrical, fire or mechanical hazards during intended generator operation. It also includes a section on testing for determination of output power rating that it delineates as non-safety based. PGMA G300 also includes manufacturing tests to ensure minimum levels of safety for production units. Although the standard restates the mandatory CPSC label requirement for portable generators, it does not otherwise address the risks related to CO poisoning.

CPSC staff continues to work with PGMA and urge them to address the CO hazard.

CPSC staff participated in a PGMA technical summit on March 17, 2016, and reaffirmed this commitment.

In April 2016, PGMA informed staff that “the PGMA Technical Committee will create a performance based standard that addresses the CO hazard created when portable generators are misused by operating them in or near occupied spaces as its top priority. The performance standard, once developed, will be proposed to the canvass group for addition to ANSI/PGMA G300 in the next revision cycle.”

CPSC staff responded to PGMA

and met with PGMA again at PGMA's request in August

and September 2016.

Letter from PGMA to Joel Recht, dated April 20, 2016, available online at:

http://www.cpsc.gov/Global/Regulations-Laws-and-Standards/Voluntary-Standards/Voluntary-Standards-Reports/PGMALettertoRechtCPSCCooperationFinal.pdf.

CPSC staff presentation,

CPSC Staff Technical Research to Address the Carbon Monoxide Hazard for Portable Generators,

March 17, 2016.

The Commission's understanding is that PGMA's revision cycle is every 5 years.

Recht, Joel, Letter to Susan Orenga, Response to PGMA Letter to Joel Recht dated April 20, 2016, May 13, 2016.

http://www.cpsc.gov/Global/Regulations-Laws-and-Standards/Voluntary-Standards/Portable-Generators/CPSCRechtLettertoPGMAMay132016inresponsetoPGMAletterdatedApril202016.pdf.

Smith, Timothy, Log of Meeting, CPSC Staff, PGMA, and Exponent, August 12, 2016, available online at:

https://www.cpsc.gov/s3fs-public/Meeting%20Log%20for%20meeting%20with%20PGMA%202016-08-12_0.pdf.

Recht, Joel, Log of Meeting, CPSC Staff and PGMA, September 6, 2016, available online at:

https://www.cpsc.gov/s3fs-public/09%2006%2016%20Meeting%20with%20PGMA%20Follow%20up%20on%20Technical%20Summit%20on%20Carbon%20Monoxide%20Hazard%20Mitigation%20for%20Portable%20Generators.pdf.

On September 19, 2016, PGMA emailed a letter to Chairman Kaye indicating that PGMA is in the process of re-opening G300 and announcing its intent to develop a “performance strategy focused on CO concentrations.”

In the letter to Chairman Kaye and in CPSC staff's September meeting with PGMA, PGMA described only broad generalities of a framework for modifying G300 that involves testing a generator in an

enclosed space (test chamber).

The Commission looks forward to working with PGMA on developing a performance requirement addressing the CO poisoning hazard associated with portable generators. Given that PGMA described only broad generalities to CPSC regarding PGMA's intent to modify G300, the Commission does not have an adequate basis to determine if modifications to the voluntary standard would likely eliminate or reduce the risk of injury or death. In addition, because the Commission is unaware of any portable generator that is or has been certified to G300, it is unlikely there would be substantial compliance if CO emissions requirements were incorporated.

Letter from PGMA to Chairman Elliot Kaye, dated September 16, 2016, available online at:

https://www.cpsc.gov/s3fs-public/PGMALtrChairKayeVoluntaryStandardFinal.pdf.

Product Safety Letter,

PGMA Talks Broad Strokes on Standards Work with CPSC,

Volume 45, Issue 34, September 12, 2016.

3. ISO 8528-13:2016

ISO 8528-13:2016

Reciprocating Internal Combustion Engine Driven Alternating Current Generating Sets—Part 13: Safety,

is a standard applicable to portable generators sold overseas. Its requirements regarding the CO poisoning hazard are limited to labels and markings. It requires that the generating set must have a visible, legible, and indelible label that instructs the user: “exhaust gas is poisonous, do not operate in an unventilated area.” The standard also requires that the general safety information section of the instruction manual mention: “Engine exhaust gases are toxic. Do not operate in unventilated rooms. When installed in ventilated rooms, additional requirements for fire and explosion shall be observed.”

C. Adequacy of the Voluntary Standards for Portable Generators in Addressing CO Deaths and Injuries

The Commission does not believe that any of the standards discussed in the previous section are adequate because they fail to address the risk of CO hazard beyond restating the CPSC mandatory labeling requirement and the Commission does not believe that the mandatory labeling requirements, alone, are sufficient to address the hazard. Additionally, the Commission is not aware of any firms certifying products to these standards. Thus, the Commission does not believe there is substantial compliance with the standards. Therefore, the Commission concludes that the voluntary standards are not adequate in addressing CO deaths and injuries.

VIII. Response to Comments

In this section, we describe and respond to comments to the ANPR for portable generators. We present a summary of each of the commenter's topics, followed by the Commission's response. The Commission received 10 comments in response to the ANPR. Subsequently, in a two-part technology demonstration, CPSC contracted with UA to conduct a generator prototype development and durability demonstration program and contracted with NIST to conduct comparative testing of an unmodified carbureted generator and prototype generators in an attached garage of a test house facility. CPSC staff published a report regarding the results of the two-part technology demonstration program that included both the UA development and durability program and the NIST comparative testing program

and received 12 comments in response to this report. NIST published a report concerning its comparative testing of generators,

and staff received four comments in response to its report. The Commission responds to these comments, as well. The comments can be viewed on:

www.regulations.gov,

by searching under the docket number of the ANPR, CPSC-2006-0057.

Buyer, Janet,

Technology Demonstration Of A Prototype Low Carbon Monoxide Emission Portable Generator,

September 2012. (available online at:

http://www.cpsc.gov/PageFiles/129846/portgen.pdf

and in

www.regulations.gov

in docket identification CPSC-2006-0057-0002.).

Emmerich, S.J., A. Persily, and L. Wang,

Modeling and Measuring the Effects of Portable Gasoline Powered Generator Exhaust on Indoor Carbon Monoxide Level

(NIST Technical Note 1781), National Institute of Standards and Technology, Gaithersburg, MD, February 2013. (available online at:

http://www.cpsc.gov/Global/Research-and-Statistics/Technical-Reports/Home/Portable-Generators/PortableGenerators041213.pdf.

A. Mandatory Carbon Monoxide Label

Comment:

One commenter claimed that the CO hazard will continue to exist even if the Commission's demonstrated technology of the prototype were applied to commercially available generators and that “educating owners about the proper use of their generators will therefore remain the first line of defense.” The commenter claimed that, for this reason, the CPSC should “conduct a study that includes a human factors analysis to determine the effectiveness of the CPSC mandated CO warning adopted in 2007.” The commenter also encouraged CPSC to revise the mandated warning “to incorporate the standards and format” in ANSI Z535.3-2011,

American National Standard Criteria for Safety Symbols,

and Z535.4-2011,

American National Standard Product Safety Signs and Labels.

Response:

Although the Commission concurs with the commenter that the CO hazard associated with portable generators will continue to exist to some degree, even if CPSC's demonstrated technology were applied to commercially available generators, it does not necessarily follow that educating owners about the proper use of generators is, should be, or would remain, the first line of defense. Human factors and safety literature identify a classic hierarchy of approaches to control hazards, based primarily on the effectiveness of each approach in eliminating or reducing exposure to the hazard. The use of hazard communications such as warning labels is universally recognized as less effective than designing-out the hazard of the product or guarding the consumer from the hazard. Thus, hazard communications are lower in this “hazard control hierarchy” than these other two approaches.

Hazard communications are less effective because they do not prevent consumer exposure to the hazard; instead, they must persuade consumers, who see and understand the communication, to alter their behavior in some way to avoid the hazard. Thus, hazard communications should be thought of as “last resort” measures that supplement, rather than replace, product redesign or guarding efforts to address residual risks, unless these higher level hazard-control efforts are unfeasible.

Laughery, K.R., & Wogalter, M.S. (2011). The Hazard Control Hierarchy and its Utility in Safety Decisions about Consumer Products. In W. Karwowski, M.M. Soares, & N.A. Stanton (Eds.),

Human Factors and Ergonomics in Consumer Product Design: Uses and Applications

(pp. 33-39). Boca Raton, FL: CRC Press; Vredenburgh, A.G., & Zackowitz, I.B. (2005). Human Factors Issues to Be Considered by Product Liability Experts. In Y.I. Noy & W. Karwowski (Eds.),

Handbook of Human Factors in Litigation

(Chapter 26). Boca Raton, FL: CRC Press; Williams, D.J., & Noyes, J.M. (2011). Reducing the Risk to Consumers: Implications for Designing Safe Consumer Products. In W. Karwowski, M.M. Soares, & N.A. Stanton (Eds.),

Human Factors and Ergonomics in Consumer Product Design: Uses and Applications

(pp. 3-21). Boca Raton, FL: CRC Press.

The commenter recommends that CPSC conduct a study to determine the effectiveness of the CPSC-mandated CO warning. The commenter states that testing is needed because of the importance of “educating owners about the proper use of their generators.” Based on this assertion, the Commission infers that the commenter's measure of effectiveness is the extent to which the warning is understood by consumers, assuming the warning had initially captured and maintained the

consumers' attention. CPSC's mandatory labeling requirements for portable generators states that the product label shall be located on a part of the generator that is “prominent and conspicuous to the operator,” while performing at least two of the following operations: Filling the fuel tank, accessing the receptacle panel, and starting the engine.

The rule also requires that the label remain permanently affixed, intact, legible, and largely unfaded over the life of the product.

These requirements, as well as the minimum type size requirements,

were developed purposefully to address issues related to capturing and maintaining consumer attention and should address most concerns of this type, except for cases in which the user of the generator is not literate in English. However, the question of whether the label also should be provided in languages other than English was raised and addressed in detail in the final rule.

In summary: (1) Available generator-related incident data have revealed no pattern of incidents involving people who could not read English; (2) the overall positive impact of adding another language to a label is likely to be very small; and (3) the regulation does not prohibit the addition of another language version of the warning message to the mandatory label.

16 CFR 1407.3(a)(iii)(B).

16 CFR 1407.3(a)(iv).

The signal word “DANGER” must be in letters not less than 0.15 inches and the remaining text must be in type whose uppercase letters are not less than 0.10 inches, or about 10-point type size.

72 FR 1443 (January 12, 2007).

The Commission supports the testing of warnings and other hazard communications. However, as discussed in the preamble to the mandatory labeling final rule, an independent contractor already performed focus-group testing with low-literacy individuals on the product label initially proposed in the notice of proposed rulemaking (NPR), and the Commission revised the final label to address the message text comprehension problems identified during testing.

The Commission acknowledges that incremental improvements to the language of the label might be possible by conducting additional comprehension testing. However, the Commission also believes that the most significant label comprehension problems have already been addressed and that additional testing of this sort is unlikely to detect problems that would substantially impact comprehension among those at risk.

In terms of the formatting of the mandatory label, the Commission notes that the formatting and requirements of the mandatory generator label are virtually identical to the requirements of ANSI Z535.4-2011 and Z535.3-2011. Although the Commission acknowledges that the formatting of the mandatory label technically does not match the panel format requirements of ANSI Z535.4, these differences were deliberate and intended to improve warning comprehension. In addition, the Z535 series of standards includes exceptions and examples that are consistent with the formatting of the mandatory label. Revising the mandatory label to strictly meet the panel format requirements of Z535.4 is unlikely to improve the effectiveness of the label, and the Commission believes such changes actually could have a negative impact because it would separate the graphics from the relevant safety messages. Thus, the Commission believes that such revisions are neither appropriate, nor desirable.

Id.

Virzi, R.A. (1992). Refining the test phase of usability evaluation: How many subjects is enough?

Human Factors, 34

(4), 457-468, has found that about 80 percent of all usability problems tend to be detected with only four or five subjects; about 95 percent of all problems are detected with nine subjects; and each additional subject was less likely to detect new usability problems. The Commission believes that these general principles are likely to apply to comprehension testing as well, particularly in tests that oversample low-literacy individuals.

B. Technical Requirements/Specifications

Comment:

Two commenters state that significant engine design changes would be required to incorporate and adapt emission technologies for use into any prototype portable generators. The commenters assert that engine designs that incorporate the prototype design changes are possible, but may not be suitable for all engines, especially when considering price and reliability considerations.

Response:

To reduce the CO exhaust levels in portable generator units, staff developed the prototype generator with commercially available parts for better fuel delivery controls and exhaust emission controls. The prototype generator did not require extensive design changes. The prototype generator engine was derived from a readily available unit with a carburetor-equipped engine, which was retrofitted with sensors and components for electronic microprocessor controls of the intake manifold fuel injection and combustion spark timing. The prototype engine with electronic fuel controls required no disassembly between the engine cover, engine block, or cylinder head. Therefore, the head gasket and cylinder compression rings were left in their original condition. Considering price, staff agrees that there is an added cost to EFI engines, as discussed in the preliminary regulatory analysis. As to reliability, staff notes that the prototype generator was successfully tested for its longevity in service (durability) for 500 hours, which was the rated useful life, as established by the manufacturer.

Staff notes that the CPSC prototype generator was meant to be a durability program demonstration to support substantially reduced CO emission rates and encourage research on an approach to mitigate the risk of fatal and severe CO poisoning. The prototype portable generator was not intended to be a production unit, as manufacturers would need to consider appropriate suitable designs for their engine families in portable generators. Staff's prototype findings have since been repeated by others who patterned their reduced CO emissions prototype generators on the design concept developed for CPSC by the University of Alabama.

See Techtronic Industries (TTi) presentation3/17/16 at PGMA's Technical Summit on Carbon Monoxide Hazard Mitigation for Portable Generators—pages 85-105 of 178 page pdf file at:

http://www.cpsc.gov//Global/Newsroom/FOIA/Meeting%20Logs/2016/MeetingLogPGMA31716.pdf.

Comment:

The Truck and Engine Manufacturers Association (EMA) asserts that similar engine designs, including basic fuel-injection and ignition design are uniform across several manufacturers' product lines of gasoline-fueled engines, where possible. Products like lawn mowers and portable generators may use a similar engine design and components, and EMA states that this uniformity across many products provides manufacturing flexibility and economy of scale. EMA states the implementation of a different engine design in portable generators, such as described in the prototype program, may impact cost and availability of the product.

Response:

The prototype design was specifically originated and developed through available off-the-shelf electronic fuel controller and components adapted onto an existing marketed portable generator engine. The prototype generator was successfully tested for its longevity in service (durability) for 500 hours, which were the longevity and emission outcomes of the new EFI engine through the rated useful life, as established by the manufacturer.

CPSC staff acknowledges the EMA concern that adoption of a portable generator engine, specifically designed to reduce CO emissions, may have

different engine components pricing compared to the current portable generator engine without the emission reduction. CPSC staff notes that portable generators with EFI (though not specifically designed for low CO emissions) have been increasing in availability in the market as new models have been introduced.

Comment:

Honda states that the photos of the prototype unit cylinder head in the University of Alabama report,

Prototype Low Carbon Monoxide Emission Portable Generator Build Description and Performance Evaluation,

may indicate that combustion gases had been leaking to the outside because the head gasket was in the early stages of failure prior to the time that the engine was disassembled. Honda indicates that they made these findings based on the carbon deposits on the prototype cylinder head fin and head gasket seating surface, shown in the photos in Figure 22 of UA's report.

UA's report (Puzinauskas, P, Dantuluri, R, Haskew, T, Smelser, J, . Prototype

Low Carbon Monoxide Emission Portable Generator Build Description and Performance Evaluation,,

The University of Alabama, Tuscaloosa, AL, July 2011) is available as TAB G in the staff report referenced previously (Buyer, Janet,

Technology Demonstration Of A Prototype Low Carbon Monoxide Emission Portable Generator,

September 2012.)

Response:

The cylinder heads, pistons and several other components are photographed and compared in the post durability wear analysis section of Contractor University of Alabama's report,

Low Carbon Monoxide Emission Prototype Portable Generator Build Description and Performance Evaluation.

Figure 22 in UA's report shows a side-by-side comparison of the cylinder heads from the baseline generator (an unmodified unit) to the prototype generator unit after completion of the 500 hours of durability testing. CPSC staff partly agrees with the Honda photo assessment because more carbon deposits are visible on the prototype cylinder head gasket surface, compared to the same component in the baseline. However, the prototype's head gasket endured approximately 585 engine hours of the durability program and subsequent emission testing. According to UA's report, the head gasket with the baseline unit leaked after 175 engine hours into the durability test and was replaced. The cylinder head photos, which compared the generator units after completion of the durability test, showing less carbon deposit on the baseline engine's cylinder head gasket seating surface may be explained by fewer accumulated engine hours on the newer head gasket. Furthermore, staff notes that the prototype engine had been run for 585 hours by the time the photograph was taken, which was 85 hours beyond the manufacturer's rated useful life of the engine.

Comment:

Honda states that that the increased combustion temperature due to the prototype's stoichiometric air-to-fuel mixture and reliance on radiant cooling is insufficient, as evident in the condition of photographed engine components, such as the pistons, after completion of the durability test.

Response:

CPSC staff agrees with Honda that leaner fuel ratios generally result in increases in combustion temperatures. Increasing the air-to-fuel ratio available for combustion was intentional in the prototype engine, to influence and reduce the CO mass flow in the exhaust emission. Cylinder head temperatures were measured in generator units at all various load profiles for each occurrence of emission testing. These emission tests occurred before modifications to engine or durability testing, during the durability testing, in which hours of engine operation were accumulated, and after the durability tests.

Emission and engine test data were collected on the as-received, carburetor-fueled generators units. According to the University of Alabama report,

Low Carbon Monoxide Emission Prototype Portable Generator Build Description and Performance Evaluation,

the as-received generator unit selected to become the prototype, but not yet modified, measured a 13.98 AFR at full generator loading (mode 1), with an associated 227 °C cylinder head temperature. In addition, the range of AFR values for this pre-modified prototype generator measured 13.98-11.26, with progressively richer AFRs toward idle or no-load. The maximum cylinder head temperatures with the stoichiometric EFI after prototype engine modification were no hotter than the original unit. Staff believes that the 14.0 AFR carburetor design offered no cylinder head cooling capacity over the stoichiometric EFI design. Throughout the prototype generator program, including independent laboratory dynamometer emission testing after 500-cyclic engine hours of operation, the engine demonstrated a cylinder head temperature less than 227 °C at full load. The mid-to-no load operating temperatures were cooler. All of these recorded measurements of the prototype cylinder head temperatures, including full load, were well below the manufacturer-recommended temperature limits.

Another comparison of cylinder head temperatures involves the baseline generator, which remained unmodified as the original unit, and the prototype generator. According to the

Low Carbon Monoxide Emission Prototype Portable Generator Build Description and Performance Evaluation

report, the carburetor fuel system of the baseline generator delivered 13.4 to 10.5 AFR values for the range of generator loads throughout the durability program. Similar to the pre-modified prototype generator, progressively richer AFRs occurred in the baseline generator towards idle or no-load. Alternatively, the prototype generator fuel strategy sought to maintain the same stoichiometric AFR across all loads. These differences in AFR values created an average elevated temperature of 28 °C in the prototype unit to the baseline unit. Staff believes the 28 °C average hotter temperatures across all loads created more discoloration in the prototype piston. There appears to be more blackened areas of the piston ring, and more coloring below the seated position of the piston ring indicate hotter operating temperatures in the prototype cylinder compared to the baseline unit. However, as mentioned, the recorded measurements of the prototype cylinder head temperatures, including full load, were well below the manufacturer recommended temperature limits. For the technology demonstration program, the prototype's leaner AFR to minimize CO exhaust production was believed to be balanced with higher, but acceptable, cylinder temperatures.

Comment:

EMA states that greater CO emission levels occurred with the prototype portable generator at 500 hour end-of-life compared to zero hour, suggesting that some deterioration of the prototype engine occurred with accumulation of engine hours.

Response:

The UA report contains an appendix with prototype and baseline generator engine-hour durability emission test results for low-, high- and mid-life engine hours. This appendix shows prototype portable generator post-catalyst CO emission results at 2 g/kW-hr near 0 engine-hours and 17.5 g/kW-hr at 500 engine-hours. Staff does not believe that these results reflect deterioration, but rather, a mid-load controller calibration performance issue, which surfaced primarily in the post-durability emission tests.

This 500-hour prototype emission test performance was due to portions of the fuel look-up tables

that were not

calibrated in the initiation of the engine build. Initially, it was not known that rated engine speeds supporting an alternator would involve extensive variation. Therefore, only certain areas of the controller look-up tables were mapped. Retrospectively, it is known that the mode 4 or mid-load solution was simply to expand the same parameters throughout the ECU look-up tables and all engine speeds. In the final emission tests, larger AFR excursions and higher CO emissions occurred when the engine operated in the unmapped portions of the controller. While the post durability prototype generator CO emissions results show more than 90 percent reduction over the baseline unit, the emission reduction with the prototype could likely be reduced further with more comprehensive calibration of the controller.

The fuel look-up tables are part of the electronic programming of the Engine Control Unit

(ECU) of the EFI system. The tables are used to associate engine operating parameters measured by the system's sensors with how much fuel the injectors need to deliver to the combustion chamber in order for the EFI system to maintain the desired air/fuel mixture.

Comment:

Honda states that the CPSC testing did not evaluate engine and generator performance in transient load conditions of performance.

Response:

The empirical testing in the NIST test house included transient loads. NIST Technical Note 1781,

Modeling and Measuring the Effects of Portable Gasoline Powered Generator Exhaust on Indoor Carbon Monoxide Level,

describes how NIST evaluated the performance of both the prototype and baseline unmodified generators in the garage, with several electrical loading variations, including the generator cyclic load profile in the durability program and emission testing.

The measuring test equipment at the NIST test house continuously collects CO measurements as the electrical and engine load profile was altered. The proposed performance requirement is based on measuring emissions while the generator is operating with a steady load applied, as opposed to a transient load.

Emmerich, S. J., A. Persily, and L. Wang,

Modeling and Measuring the Effects of Portable Gasoline Powered Generator Exhaust on Indoor Carbon Monoxide Level

(NIST Technical Note 1781), National Institute of Standards and Technology, Gaithersburg, MD, February 2013 (available online at:

http://www.cpsc.gov/Global/Research-and-Statistics/Technical-Reports/Home/Portable-Generators/PortableGenerators041213.pdf

and in:

www.regultations.gov

in docket identification CPSC-2006-0057-0005.)

Comment:

Two commenters asserted that CPSC's prototype components may cause exacerbated reliability issues after long-term storage.

Response:

Staff disagrees because fuel-injection improves reliability. A fuel-injected system is sealed, so the fuel is not exposed to air like the vented system associated with a carburetor. Exposure to air significantly contributes to degrading gasoline during long-term storage and, in turn, causes problems with starting and running the engine. Manufacturers advertise improved reliability as one of the benefits associated with fuel injection.

Comment:

One commenter asserted that it is harder to apply EFI and catalyst on the smaller engines used in 1 kW-3 kW units and that they are sold in higher numbers than 5 kW units. In a similar comment, another commenter noted that CPSC's prototype used a commercial-grade engine in open frame, yet closed-frame units are more popular.

Response:

CPSC has observed that there are fuel-injected handheld Class I engines, with and without catalysts, in the marketplace. CPSC acknowledges, however, that there may be more challenges associated with implementing the emission control technology on these smaller engines and the generators that these engines power. Thus, there is a later compliance date in the proposed rule for these models, relative to the larger generators powered by Class II engines. Based on CPSC staff's analysis of the market data, CPSC concurs that smaller generators are becoming more popular, relative to larger generators. CPSC staff used a larger generator, powered by a class II, single-cylinder engine, in the technology demonstration program because the Commission's incident data show that generators with these engines were associated with almost two-thirds of the CO deaths involving generators that have been reported to CPSC, when the size of the generator was identified, for the years from 2004 through 2012. The lower proposed performance requirements for smaller generators are expected to reduce deaths that could otherwise be expected to occur with increasing popularity of these smaller units.

Comment:

One commenter stated that stable engine operation under transient loads requires richer-than-stoichiometric AFR. Without it, the commenter asserted, there is unreliable operation, which can result in damaged electrical loads and warranty claims.

Response:

The Commission acknowledges this operating challenge, and for this reason, the proposed performance requirement is based on measuring emissions while the generator is operating with a steady load applied, as opposed to a transient load.

Comment:

One commenter noted that their company uses more severe modes and requirements to test product durability, which they are doubtful the prototype would have survived. In a related comment, a commenter asserted that significantly reduced CO emissions at the highest loads resulting from operation near stoichiometric fuel control will negatively impact engine durability.

Response:

The Commission notes that the proposed performance requirement for generators powered by class II single-cylinder engines is nominally six times higher (less stringent) than the CO rate that the prototype generator achieved. The Commission believes that the proposed CO emission requirements can be achieved on many existing engines by replacing the carburetor with closed-loop EFI and integrating a catalyst without engine design modification and without negatively impacting engine durability. The Commission notes, however, that for some engines, modifications might be needed to enable operation closer to stoichiometry. For other engines that cannot be improved through design modifications, those could still be used in generator applications by using a product integration strategy that precludes installed engine operation at loads where fuel enrichment is needed.

10.

Comment:

One commenter stated that the performance standard for CO emission rates must take into account deterioration of emissions to achieve the target exposure over the life of the engine.

Response:

The Commission took deterioration into account in developing the performance requirements. The Commission believes deterioration of CO emissions to be minimal. This is based on both the performance of CPSC's durability-tested prototype at end of life as measured by CES, as well as by observation of published deterioration factors for CO, which are measures of the increase in CO emissions for an aged engine, relative to its emissions when new. The Commission observed in the EPA's exhaust emission database for model year 2015 that a vast majority of the engines have a deterioration factor below 1.1 (thus indicating the emissions worsen by less than 10 percent above initial emissions).

11.

Comment:

One commenter stated that the target CO emission rate in terms of g/kW-hr should be based on engine displacement, with lower rates (in terms of g/kW-hr) for larger engines to achieve the same target exposure.

Response:

The Commission believes lower CO emission rates are technically feasible for smaller engines, compared to larger engines. Consequently, the Commission is proposing performance requirements for four different size

categories of generators that are each based on technical feasibility and analysis of benefits and costs as a function of engine displacement, and, for the largest category, also whether the engine has one or two cylinders. The epidemiological benefits considered exposure differences for different generator types, by allocating known incidents based on location of generator and location of victims in various house types.

12.

Comment:

One commenter asserted that reducing CO emissions will increase other pollutant emissions and risk of fire and burn hazard.

Response:

The Commission does not agree that reducing CO emissions will increase other pollutant emissions. Based on the emission results from CPSC's prototype generator, as well as those from the EPA's demonstration program, reducing CO emission rates also results in reduced HC+NO

emissions. CPSC staff acknowledges that for CPSC's prototype, the leaner air fuel ratio resulted in elevated exhaust temperatures compared to the carbureted configuration. Staff notes, however, that the muffler that was used was chosen to easily accommodate integration of the small catalyst into it. This muffler had less internal baffling, which resulted in average muffler surface temperatures of approximately 70°C hotter than the OEM design. As a result, UA shrouded this muffler and that resulted in shroud surface temperatures that were lower than the OEM muffler that was not shrouded. Staff notes that use of better designed mufflers, and, if needed, improved flow of cooling air over the exhaust, could mitigate the effect of elevated exhaust temperatures.

13.

Comment:

One commenter stated that EFI systems are becoming more low cost and noted that an oxygen sensor of one particular design can serve as a safety switch if the engine starts operating rich of stoichiometric.

Response:

The Commission has observed that small SI engines with EFI have entered the marketplace in recent years, and expects this would mean that they have become less expensive. The Commission is interested in combining reduced CO emissions with a mechanism that will shut off a generator when operated in an enclosed or semi-enclosed space.

14.

Comment:

One commenter stated that the results from testing the generators in NIST's garage should not be relied upon for any rulemaking related to portable generator safety because, the commenter asserted, the attached garage on NIST's test house is not sufficiently representative of how garages are conventionally constructed.

Response:

The Commission used the results from NIST's test house to provide an example of the reduction in the house's hypothetical occupants' exposure that the reduced CO emission rate from a portable generator can yield when compared to a current carbureted generator when operated in the same garage. The Commission is basing the proposed performance requirements for the rule on technically feasible CO emission rates, along with an assessment of the impact of those rates through indoor air quality modeling of 40 structures, representative of the U.S. housing stock, where generators were operated in 503 of the deaths in CPSC's databases that occurred from 2004 through 2012.

15.

Comment:

Several commenters expressed concern about CO deaths caused by generators and expressed support for reducing generators' CO emission rates and their belief in the technical feasibility to do so.

Response:

The Commission agrees with the commenters.

C. CO Poisoning Effects

Comment:

The commenter considers that CPSC staff assumes COHb levels below 10 percent are not harmful. The commenter notes that there is no scientific basis for such an assumption and also notes that, in many studies, COHb levels do not correlate consistently with symptoms.

Response:

The Commission does not assume that a CO exposure resulting in less than 10 percent COHb is incapable of causing adverse health effects. The Commission has long recognized the existence of populations especially sensitive to CO health effects (fetuses, asthmatics, and individuals with cardiovascular diseases). Most authorities, including CPSC, consider individuals with coronary artery disease [CAD] to be the population most sensitive to potential adverse health effects of CO at the lowest exposure levels. Some studies report individuals with CAD might perceive adverse health effects, and/or, tests show that they may experience adverse health effects that they are unaware of, at about 2 percent to 5 percent COHb. The Commission understands that the pathophysiological effects of CO are complex and strongly influenced by multiple factors, particularly CO level, exposure duration, and exposed individual's inhalation rate and health status. In the ANPR on portable generators, and in the prototype report documents, CPSC focused on extremely high-level, acutely lethal, CO exposures caused by generator exhaust. Therefore, rather than provide an exhaustive review of all studies, including equivocal findings in some low-level exposure studies, CPSC is providing an overview of the complex interactions between multiple variables that influence the end effects of acute, high-severity CO exposures in humans. CPSC emphasizes that CO poisoning effects should be understood to be a continuum of effects of the exposure, rather than be viewed as discrete health effects tightly tied to specific CO levels or COHb levels.

Comment:

One commenter stated that although a low CO emissions generator would undoubtedly save lives if widely applied, “prediction of confusion and incapacitation from COHb levels is not possible.” The commenter cited his recent publication reporting that “symptoms of CO poisoning do not correlate well with COHb levels.” Based on his findings and other clinical reports, the commenter questions the validity and/or concept of a table relating COHb levels to particular symptoms, as used by the Commission. The commenter believes that it is incorrect to use COHb levels to calculate egress times from a CO-containing environment and notes that there are no data to support the method. Another commenter also questioned the validity of an approximate relationship between COHb levels and severity of CO poisoning symptoms and health effects.

Response:

The Commission's use of predicted COHb levels was not intended to calculate an actual egress time from a CO exposure, and the Commission noted that reduced emission generators would not guarantee egress by exposed individuals. Rather, the Commission considers that reduced generator CO emissions, as achieved with its prototype unit, will substantially delay the rate at which CO levels rise in poorly ventilated spaces, and will thus delay the rate at which COHb levels of exposed individuals rise (in some cases reducing the peak COHb level attained). This will provide significantly increased time available for individuals to remove themselves from the exposure environment or to be rescued by an outside party. Supporting evidence that some individuals will react appropriately to slower onset of CO poisoning effects has been reported (

e.g.,

111 of 167 patients with CO poisoning presented to Florida hospital emergency departments (ED) between 5 a.m. and 10 a.m., after waking and feeling ill consequent to overnight use of a generator during hurricane-related power outages). CPSC data indicate that in 69 of 93 cases where it was known

how and why a patient with generator-related CO exposure presented to an ED, the patient had either transported themselves or contacted others (9-1-1, family, friends) to arrange for their transport to the ED. In the remaining cases, individuals were found in distress by others (either a lesser affected co-exposed individual or an outside party).

The Commission recognizes that even healthy individuals can exhibit variability in individual susceptibility to CO health effects under identical exposure scenarios. The Commission understands that, in clinical situations, CO poisoning symptoms and health effects do not necessarily correlate well with a patient's initial COHb measurement, which is often confounded (generally reduced by factors such as time interval relative to cessation of CO exposure and provision of supplemental oxygen). Clearly, COHb measurements can be of limited value to physicians when determining appropriate treatment plans for individual patients. Rather than make clinical decisions, the Commission needed to provide controlled, systematic comparisons of how CPSC's reduced CO emissions prototype generator could be expected to reduce the lethal CO hazard presented by the unmodified original generator. Therefore, CPSC used identical physiological input parameters for a healthy adult to model COHb formation and elimination from empirical generator CO time course exposure data. CPSC used predicted times taken to rise to, and progress through, three convenience benchmark percentile COHb values to compare the relative CO poisoning hazard presented by a generator before and after design modifications to reduce its CO emission rate. The Commission considered these benchmark values to approximate relatively mild (20% COHb), potentially incapacitating (40% COHb), and likely lethal (60% COHb) exposure levels. Although indicating health effects generally first reported at these benchmark COHb levels, CPSC did not intend to convey that they represented precise measures when appearance of symptoms and adverse health effects would be expected in all individuals. CPSC noted that rapidly rising, high-level CO exposures of several thousand ppm (as can occur with current carbureted generators) would result in extreme oxygen deprivation and fast-rising COHb levels, causing rapid incapacitation, loss of consciousness and death, without individuals necessarily experiencing milder, progressively worsening CO poisoning symptoms typically manifested in slowly rising or lower-level CO exposures.

As further detailed in the staff's briefing package, the available physiological research data and clinical findings in the scientific literature support the use of “COHb benchmarks,” for approximate estimation and comparison of CO-related health effects expected during generator-related exposures.

The Commission welcomes suggestions on alternative health-based approaches to compare the reduced CO emissions generators with current products in terms of improved safety benefits.

Tab K, Appendices, of staff's briefing package.

D. Jurisdiction

Comment:

One commenter asserted that pursuant to § 31 of the CPSA, the CPSC lacks authority to regulate the risk of injury associated with CO emissions from portable generators because that risk could be addressed under the Clean Air Act (CAA). Specifically, the commenters rely on Section 213 of the CAA, which directs the EPA to conduct a study of emissions from non-road engines to determine if they cause or contribute to air pollution, “which may reasonably be anticipated to endanger public health or welfare.” 42 U.S.C. 7547(a)(1)(2006). Under this provision of the CAA, the EPA has promulgated regulations governing CO emissions from portable generators. In particular, 40 CFR part 90 imposes requirements to control emissions from non-road spark-ignition engines, which includes portable generators, at or below 19 kilowatts.

Response:

Section 31 of the CPSA does not establish an absolute prohibition to CPSC action whenever the CAA is implicated. Rather, the Commission lacks authority to regulate a risk of injury associated with a consumer product if that risk “could be eliminated or reduced to a sufficient extent through actions” taken under the CAA. 15 U.S.C. 2080(a). Case law and the legislative history of § 31 confirm this.

See ASG Industries, Inc.

Consumer Product Safety Comm'n,

593 F.2d 1323 (D.C. Cir. 1979) (under section 31, CPSC is to consider all aspects of the risk and make a judgment whether the alternate statute can sufficiently reduce the risk of injury).

The legislative history indicates that Congress contemplated a stricter ban on the CPSC's jurisdiction and rejected it. Specifically, the Senate version of the bill for § 31 would have precluded CPSC's jurisdiction if the product was “subject to safety regulations” under one of the statutes listed in section 31 of the CPSA. S. Rep. No. 92-749, 92d Cong., 2d Sess. 12-13 (1972). In contrast, as the

ASG

court noted, under the House version of the bill, which was eventually enacted, the Commission has authority if there has not been sufficient reduction or elimination of the risk of injury. H.R. Rep. No. 92-1593, 92d Cong., 2d Sess. 38 (1972).

The CAA and the EPA regulations promulgated under it that address CO emissions from portable generators have not sufficiently reduced or eliminated the risk of CO poisoning associated with portable generators that the CPSC seeks to address. Deaths and injuries associated with CO emissions from portable generators have increased since the EPA adopted its regulations limiting CO emissions from the type of engines used in portable generators.

The CAA and the EPA's regulations create national standards intended to address large-scale ambient air pollution, not acute CO exposure from portable generators. The CAA and the EPA's regulations, created under 42 U.S.C. 7407, are designed to reduce CO emissions in regional areas that exceed National Ambient Air Quality Standards. These requirements are not designed to reduce the localized risk to consumers from acute CO poisoning when portable generators are used in the home.

Additionally, EPA's 2008 adoption of an averaging program for CO emissions from marine engines further demonstrates that its regulations are not concerned with the risk of acute CO poisoning, but only large-scale overall emission levels. This averaging program allows a manufacturer to exceed the EPA's CO emission limits for a group of similar engines, as long as the manufacturer offsets that increase with another “engine family” with emission levels below the EPA's limit. 73 FR 59,034 (Oct. 8, 2008). It is noteworthy that this averaging program applies to CO emissions from marine engines, which the EPA explicitly acknowledges are associated with “a substantial number of CO poisonings and deaths.” 73 FR 59,034, 59,048 (Oct. 8, 2008). Under this program, emissions from an individual engine are inconsequential to EPA's rule, and so is the individual consumer's exposure level. Rather, the EPA's determination of CO emission limits focuses on ambient air pollution on a large scale.

Finally, the structure of the CAA and its delegations of authority make the EPA unable to adequately address the risk of injury associated with CO poisoning to consumers from portable

generators. Under the CAA, the EPA sets National Ambient Air Quality Standards (NAAQS) and has oversight and enforcement authority, but the states retain primary responsibility for ensuring air quality. Section 107 of the CAA sets out states' responsibilities for ensuring air quality, including determining how the state will meet NAAQS, and identifying attainment and non-attainment areas. 42 U.S.C. 7407. The U.S. Supreme Court has emphasized that the EPA is “relegated by the [CAA] to a secondary role,” as long as states adopt plans that meet the general requirements.

Train

Natural Resources Defense Council, Inc.,

421 U.S. 60 (1975). This broad leeway provided to states indicates that the CAA and the EPA's regulations are not intended to and cannot provide sufficient specificity to mitigate the risk of CO poisoning.

E. CO Sensor Systems and Exhaust Pipe Extension

1. Generator-Mounted CO Sensing Shutoff Systems

Comment:

Four comments were submitted on the concept of a generator-mounted safety shutoff system using CO sensing technologies that could be used to limit consumer exposure to CO present in portable generator exhaust. Three of the four commenters advocated for such a system, and one advocated against it.

One comment in support of the use of residential CO alarm technology noted that a CO sensor that is used to activate ventilation systems in parking garages can be used for turning off the generator when it senses 35 ppm CO. The Commenter also recommended that the system be interlocked to prevent generator operation every 2 to 3 years, when the sensor's useful life is expended, and to prevent operation, if the user disables the system.

The commenter who did not recommend the use of residential CO alarm technology expressed the belief that COS sensing technology near a generator may impair its operation, causing users to disconnect the sensors to ensure a steady source of electricity. The Commenter also noted that CO sensors require routine maintenance, and their capabilities can degrade with time and during extended periods of inactivity, adding that it may be unreasonable to expect consumers to regularly check and maintain the CO sensing equipment, particularly when the generator is not even being used.

Response:

The Commission shares the concern that using CO sensing technology in the vicinity of a portable generator may impair the generator's operation, causing users to disconnect the sensors. The Commission agrees that it is unreasonable to expect consumers to regularly check and maintain CO sensing equipment, particularly when a generator is not being used. Early in the portable generator project, the Commission investigated one version of the concept of an on-board CO sensing shutoff system; the investigation and its findings are documented in the staff report,

Phase 2 Test Report: Portable Generator Equipped with a Safety Shutoff Device

(Brown, 2013). Its goals were to: (1) Determine if a CO sensor/alarm output signal from commercially available residential CO alarms (meeting the requirements in UL 2034

Single and Multiple Carbon Monoxide Alarms

), when retrofitted with circuitry connected to the generator, could trigger a shutoff device installed on a portable generator when the CO alarm activated; and (2) measure CO concentrations around the generator when operated in multiple environments to assess CO migration and levels that might occur under several scenarios. Test environments examined included outdoors, in a two-sided structure, as well as inside and under a temporary modular storage (TMS) building.

In that investigation, the Commission found that when the generator was operated inside the TMS building, the CO migrated and accumulated on the far side of the room more quickly than near the generator. The CO alarms on the generator never activated before those located elsewhere in the space activated, with the time difference generally ranging from 5 to 10 minutes. In some tests, CO levels in some parts of the room reached up to 1,000 ppm before the CO alarm on the generator activated and shut off the generator. When the generator was operated in wide-open outdoors in a light breeze condition, CO concentrations ranging up to 350 ppm were measured in the immediate vicinity of the generator. Although this did not activate the CO alarms mounted on the generator to shut it off, the Commission believes this could occur in some circumstances. This would detrimentally affect the utility of the generator when used in a proper location.

In addition to these performance deficiencies, the Commission is concerned about the ability of CO sensors to survive the environments produced by an operating generator. Currently available electrochemical and semiconductor CO sensors, which dominate the CO sensing market, have numerous vulnerabilities that will compromise their ability to maintain accuracy if they are used in an atmosphere containing high concentrations of hydrocarbons, as is present in a generator's exhaust, particularly when used in a confined space.

Regarding one commenter's recommendation to use CO sensors that turn on ventilation fans in parking garages, a recent energy efficiency study examining the performance of parking garages that have CO-sensing activated ventilation indicates that this type of system is subject to failure if not maintained on the manufacturer's recommended schedule (California Utilities Statewide Codes and Standards Team, 2011). Systems employing both electrochemical and solid state technology that were five and 12 years old, respectively, failed likely because they had not been calibrated. A properly maintained 2-year-old electrochemical sensor-equipped system performed well. The commenter suggested that to account for the referenced 2 to 3 year expected sensor life, the consumer replace the sensor at the end of the sensor's useful life. The Commission believes that it is not appropriate for consumers to be required to replace a primary safety device, let alone replace it every 2 to 3 years, when the life of the overall product is much longer. Furthermore, making the sensor replaceable makes it vulnerable to tampering. Notwithstanding the previously mentioned CO concentrations that CPSC measured around a generator operating in a proper location, the conflict between making the sensor consumer-replaceable and tamper-proof leads the Commission to conclude that currently available sensors are not likely to be effective, given the long service life of portable generators. With respect to the recommendation for a 35 ppm CO set point for an on-board sensor, CPSC measured CO concentrations in excess of 35 ppm in the immediate vicinity of the generator, while operating outdoors within 11 minutes after starting the generator (Fig C2 in Brown, 2013). A 35 ppm limit for shutoff would greatly limit the utility of portable generators when used properly.

2. Remotely Located CO-Sensing Shutoff Systems

Comment:

Two commenters raised concerns about the concept of a remotely located CO-sensing shutoff system, such as that investigated and documented in the staff report, “

Demonstration of a Remote Carbon Monoxide Sensing Automatic Shut-Off Device for Portable Generators”

(Lee, 2006). Conceptually, a remotely located

CO-sensing shutoff system would use a CO sensor located indoors to monitor for CO infiltration at that location and when it detects an unsafe CO concentration there, the sensing shut-off device would communicate with the generator to shut it off. The report presents CPSC staff's investigation of one version of such a concept, consisting of a CO alarm retrofitted with a wireless transmitter, placed by the user in an indoor location, which communicated with a wireless receiver mounted onto a portable generator operating in an attached garage. When the CO alarm activated, it energized a circuit on the generator and shut off the generator.

One commenter raised a number of behavioral and technical issues on the utility of such a system. This commenter noted that the same technical comments he made on the generator-mounted safety shutoff concept, discussed above, apply to the remote-sensing concept as well. This commenter also noted that remote-sensing technologies require consumers to take affirmative actions to properly locate sensors inside buildings and to monitor them to make sure that they continue to be operational. The commenter stated that the risk of the CO poisoning hazard would not be mitigated when consumers fail to locate or use the sensing technology properly or the detector malfunctions due to infrequent use or lack of maintenance.

Another commenter enumerated a number of concerns about the concept of a remote CO-shutoff system that included:

• Sensor performance affected by ambient conditions

• battery life

• the ability of consumers to install

• nuisance trips causing consumers to disable system

• the need to maintain proper battery charge

• ability of consumer to start generator, then remove the remote sensor to an area without CO, to allow the generator to operate.

Regarding the staff report, the commenter objected that only one model generator was included in the tests and that only a limited number of hazard scenarios were tested. The commenter provided a list of options that would need to be investigated to document remote CO-sensing device acceptability. The options include: (1) Effectiveness of the mandatory warning label; (2) effects of environmental conditions on CO dispersion in a building; (3) effect of generator load profile on CO dispersion; (4) effect of walls and building materials on the sensor's radio frequency (RF) signal to the generator; and (5) maximum distance between sensor/transmitter and the generator. Additional areas the commenter listed include: (6) Consumer's ability to reset the system in adverse conditions (darkness, storms); (7) timing of product sales (pre- or post-storm); (8) minimum component performance requirements; and (9) minimum battery requirements.

Response:

The Commission agrees that there are multiple challenges with a remote CO-shutoff concept for portable generators, including many of the challenges identified by the commenters and notes that the staff report concluded with the following:

The study was limited to proof-of-concept and did not consider issues such as life expectancy, reliability, usability, and environmental conditions. All of these factors would need to be considered in developing a remote CO detection/shut-off system for portable generators for consumer use.

In addition to having the same sensor-related concerns as those stated above in CPSC's response to the on-board CO sensing shutoff concept, CPSC has additional concerns, a primary one being that a system of this sort would need to be provided with the generator and would require the consumer to properly install the sensing devices. The consumer could easily defeat the features by operating the generator in an enclosed location and intentionally placing the sensor outdoors or other locations away from where the CO is infiltrating in order to keep the generator running. Another scenario of concern involves the user placing the CO sensor in a room where he/she thinks the CO will infiltrate, but the CO infiltrates faster in another room that the system is not monitoring. Transmitter range is another concern; if a consumer properly locates the generator outdoors at a distance far enough from the dwelling to prevent CO infiltration, the distance may render the generator inoperable if it is not within range of the sensor signal. Based on the concerns mentioned above, the Commission is not pursuing this concept as a means of reducing the CO hazard associated with portable generators.

3. Flexible Exhaust Pipe Extension

Comment:

One commenter recommended using an exhaust hose that has one end that fits over the tailpipe and a laterally expandable window fitting on the other end to direct exhaust out through a window. The commenter recommended that the hose should have an electrical circuit wired through its entire length, which plugs into the generator to prevent operation if the hose is not properly attached.

Response:

There are several drawbacks to this approach. First, if the hose must be attached for the generator to operate, then it must be attached even if the generator is correctly located away from the house. CPSC believes this is not practical. Second, the CPSC database includes fatal CO incidents where the generator was located outside the dwelling, but not so far away to prevent exhaust from entering the home through leaks or openings (Hnatov, 2015). Third, CPSC staff believes that it is unlikely that an expandable window insert can be installed in such a way as to be leak tight. Last, this system's successful use depends on the consumer's ability to properly install both the hose and the window fitting. Given these concerns, the hose extension is not a technically feasible approach to address the carbon monoxide poisoning hazard associated with engine-driven portable generators.

F. Economic Considerations

On February 12, 2007, counsel for American Honda Motor Co., Inc., Briggs & Stratton Company, and Yamaha Motor Corporation, USA (the companies), submitted comments jointly on the December 12, 2006 advance notice of proposed rulemaking (ANPR), concerning portable generators. The companies made the following comments on economic issues:

Comment:

The vast majority of consumers use their portable generators properly and safely. CPSC should give proper weight to the benefits and widespread uses of portable generators, as well as the affordability of current models.

Response:

Although the great majority of consumers might exercise proper safety precautions, improper use of the product can and does have disastrous consequences. The Commission evaluated different technologies to address the risk and has concluded that a performance standard that sets requirements that reduce CO emissions from generators is the most reliable regulatory alternative to address the risks of CO poisoning associated with portable generators. Manufacturing cost increases under the proposed rule would generally have a relatively greater impact on percentage price increases (and consumer demand) for low-price units, such as units lacking inverter technology (as discussed in the preliminary regulatory analysis section). However, the analysis finds that the

estimated benefits outweigh the costs to comply with the proposed rule.

Comment:

Staff has not provided consumer exposure data to support risk analysis of CO deaths associated with consumer use of generators.

Response:

Since the comment was filed, additional information and analysis has greatly improved the analysis of risks associated with consumer use of portable generators. The Commission's preliminary regulatory analysis has analyzed historical shipment information acquired from market research firms (Power Systems Research and Synovate), from federal data sources (the International Trade Commission and Bureau of the Census), and from individual manufacturers to estimate the numbers of portable generators in use, by engine class and other characteristics, during the period covered by CPSC staff's epidemiological benefits analysis (Hnatov, Inkster & Buyer, 2016). The new information and analysis has enabled CPSC to estimate CO poisoning risks (and societal costs) per generator in use. Additional information on product sales and use, which the industry is encouraged to provide in comments to this NPR, could further refine these estimates.

Comment:

In response to the technology demonstration report, one commenter stated that although engine designs that incorporate the report's design changes

are possible, they may not be suitable for all engines, including many used to power portable generators. This is especially true when considering the price point and reliability considerations associated with portable generators designed and sold to consumers for emergency or infrequ

This text is long and has been trimmed here. Open the source document for the complete record.

This is a copy of a public record, reproduced as it was published. It is not legal advice, and it may not be the version a court would rely on. Check the official source before you cite it.