Investigation of Current State, Local, and Tribal Facility Inventory Data Sources and Data Flows


Investigation of Current State, Local, and
Tribal Facility Inventory Data Sources and

Data Flows

1 Overview

The goal of this report is to explore how facility data workflows would have to happen between
CAERS and State, Local, and Tribal authorities (SLTs), so that CAERS may be developed to accommodate
such SLT workflows with as much flexibility as possible.

Previously, four use cases of SLT workflows had been identified at a conceptual level, depending on
whether an SLT would prefer to keep some or all their current SLT interface and backend. These cases
are:

Case 1: the SLT interface and backend are retained (CAERS receives data from SLT system) to
distribute to the federal programs.

Case 2: SLT interface and backend are retained (CAERS receives data from facilities and pushes
data to state interfaces).

Case 3: CAERS replaces SLT interface, but state databases are retained.

Case 4: SLT uses CAERS for reporting instead of their previous reporting system or method.

The current SLTs using CAERS (GA, Washington D.C., Pima County AZ, and Rhode Island) are all
considered to fall under Cases 3 or 4.

However, details about how facility data are updated and maintained by SLTs require research because,
regardless of the case the SLT falls under, SLTs may choose not to use CAERS as the primary repository
of their facility inventory data. Furthermore, there may be additional data fields that SLTs require for
their facility inventory, that would need to be accommodated in CAERS to enable SLTs to take full
advantage of combined air emissions reporting. Overall, while SLTs may focus on the uses of their
facility inventory to meet their own needs (permitting compliance, billing, emissions inventory analysis),
the ability to share the same facility inventory with the federal programs via CAERS, would keep
emissions inventory data reporting at the facility and sub-facility levels consistent across programs. This
means that each program would have a similar version of a facility in terms of facility and sub-facility
data, instead of slightly different version of the same facility which happens currently. Also, reporters
wouldn't have to report their emissions data associated to different versions of the same facility and
sub-facility components to each program. Facility and sub-facility components shared amongst SLT and
federal programs would ultimately provide data analysts, inventory developers, and modelers, the
ability to conduct multipollutant analyses across several different types of pollutants; hazardous air
pollutants (HAPS)/toxics, criteria air pollutants (CAPs), and greenhouse gases (GHGs).

To ensure that CAERS can accommodate specific facility data workflows, so that more SLTs may use
CAERS, the Facility Research & Development (R&D) Team conducted a research project to find out what

1

-------
data sources and data flows SLTs use currently for obtaining and updating facility inventory information
for their respective emission inventories. The results will help inform how CAERS would need to be
enhanced to send facility inventory data to and receive facility inventory data from SLTs.

Enhancements in CAERS could include both workflows: the ability to send data to and from a given SLT's
system, additional SLT-specific data fields that might need to be included in CAERS, as well as potential
data transformations which could be done by the SLT before sending data to CAERS, or as part of an SLT
CAERS module. An SLT CAERs module refers to a set of coding that would be added in CAERS or as part
of an SLTs workflow with CAERS, to meet SLT specific requirements (that other SLTs don't need), but
that also does not require a rebuild of CAERS.

2 Method

The R&D team gathered input about information needed to understand SLT facility inventory data
workflows from team members and the PDT. State members in the R&D team designed and finalized a
questionnaire based on that input, and ECOS sent the survey out to SLT contacts who report to the
National Emissions Inventory (NEI), in April of 2022. The questionnaire referenced the new Emissions
Inventory System (EIS) Consolidated Emissions Reporting Schema (CERS) v2.0, to ensure that
questionnaire respondents were familiar with the data fields and definitions mentioned in the survey.

CERS is used by the Emissions Inventory System (EIS) to provide a standardized structure for delegated
authorities to submit emissions inventory data to the United States Environmental Protection Agency
(EPA) to meet the Air Emissions Reporting Rule (AERR) requirements. The questionnaire used simplified
data components (Figure 1) to cover all information specified in CERS for facility inventory that is
currently being collected.

Figure 1. Simplified Representation of CERS Facility and Sub-facility Inventory Components Used in The Questionnaire

The online questionnaire used a multiple-choice format (see Appendix A). It was sent to SLT point source
contacts via e-mail on 4/5/2022. Two follow-up reminders were sent out on 4/18/2022 and 4/27/2022,
respectively. ECOS received 55 jurisdictional responses out of 78 jurisdictions that were contacted, and
one incomplete response had to be excluded, leaving 54 valid responses (a 69% response rate). Table 1
summarizes the number of SLTs responses to the questionnaire.

-------
Table 1. Number of Questionnaire Responses by Respondent Type

Respondent Type

Local

State

Tribal

Incomplete

Total

Valid

Individuals Contacted

107

Jurisdictions Contacted

| 23

Individuals Responded

Jurisdictions Responded

I 16

To clarify the responses further, SLT team members sent follow-up emails to 18 SLTs. Data clean-up was
also performed to make responses consistent. For example, one responder did not select "State
permitting program" in answering the question regarding information sources for facility sites. It
indicated the "other" choice and entered "Permittees submit annual throughput info to the Permitting
Section, which calculates the emissions and charges the Permittees an annual emission fee. The
Emissions Inventory (El) group uses the calculated emissions from the Permitting calculation sheets."
This implies the jurisdiction obtains facility site information from a permitting program. Therefore, the
choice of "State permitting program" was considered as the actual response. Another example of data
clean-up is an initial indication in the "other" choice of the use of "Inspections", which means that the
State compliance program was the right choice. In this case, the choice of State compliance program
was considered as the response.

There were 4 states for which multiple people from the state responded to the questionnaire with
inconsistent answers. These responders were contacted, and the correct answers were confirmed and
adjusted accordingly to represent one complete answer for their jurisdiction.

3 Analysis of Questionnaire Responses
3.1 Type of Facilities in Emission Inventories

All 54 valid responses indicated the inclusion of facilities with Title V permits in their emission
inventory, while 43 also included facilities with other types of permits (such as minors, synthetic minor,
etc.). Of the responses, 14 jurisdictions indicated they have non-permitted facilities in their point
sources emission inventory. Those non-permitted facilities include greenhouse gas (GHG) emission
facilities that are not subject to reporting criteria air pollutants (CAPs) by EPA's Air Emission Report
Requirements (AERR) and/or SLT requirements, facilities subject to other state requirements, TRI
facilities, small facilities with locational coordinates to support reporting for EPA nonpoint source
emission inventory requirements, such as landfills, and publicly owned treatment works (POTWs).

Table 2 and Figure 2 shows a further breakdown of types of facilities handled for emission inventories by
different jurisdictions.

This clearly indicates CAERs could be used by SLTs not only for facilities that are subject to federal-
reporting requirements but also for non-federal reporting requirements in use cases 2-4 of SLT
workflows. Otherwise, the SLTs would need to have one reporting system for facilities that also report to
one or more federal programs, and a separate reporting system for those that don't, and this may or
may not be preferred by SLTs.

-------
Table 2. Types of Facilities by Jurisdiction



Title V Permit

Other Permit

Non-Permit

Local

16

13

2

State

37

28

11

Tribal

1

1

1

Total

54

42

14

Figure 2. Number ofSLTs with Facilities in Point Source Inventories by Permit Type

60

Title V Permit	Other Permit	Non-Permit

¦ Local ¦ State ¦ Tribe

3.2 Number of Facilities in Emission Inventories

The total number of unique facilities within a given SLT point source emission inventory ranges
from tens-of-thousands (for example, 30,000 in Colorado and 28,000 in Wyoming) to less than 10 (5 in
Knox County, Tennessee). From the data collected, the median number of facilities in SLT point source
inventories is 271.

SLTs may not report to the NEI all the facilities for which they collect emissions data. The total number
of facilities reporting to the NEI by SLTs could be as many as 28,000 in Wyoming; and as few as 5 in Knox
County, TN. The median number of facilities reported to the NEI by SLTs is 149.

This information speaks to the need for CAERS to have the capacity to hold large facility inventories for
SLTs who may need them and ensure that speed and system performance is not slowed by large facility
inventories.

4

-------
3.3 Data in Tribal Emission Inventories

The U.S. government officially recognizes 574 Indian tribes in the contiguous 48 states and
Alaska. Although each tribe is different, most tribes do not have their own emission inventory and do
not report to the NEI, as many are not required to report. Tribes with Treatment as a State (TAS) for
emissions inventories are subject to AERR reporting, and tribes can voluntarily accept this responsibility
in coordination with states and EPA. Eight tribes reported to the 2017 NEI. Table 3 provides details.

Table 3. Number of Facilities Reported by Tribes for the 2017 NEI.

Tribal Name

EIS Code

Program
System Code

Number of
Facilities

Coeur d Alene Tribe of the Coeur d Alene Reservation,
Idaho

88181

TR181

Confederated Tribes and Bands of the Yakama Nation,
Washington

88124

TR124

Nez Perce Tribe of Idaho

88182

TR182

Northern Cheyenne Tribe of the Northern Cheyenne
Indian Reservation, Montana

88207

TR207

Salt River Pima-Maricopa Indian Community of the Salt
River Reservation, Arizona

88615

TR615

Shoshone-Bannock Tribes of the Fort Hall Reservation of
Idaho

88180

TR180

Southern Ute Indian Tribe

88750

TR750

293

Ute Mountain Tribe of the Ute Mountain Reservation,
Colorado, New Mexico & Utah

88751

TR751

Grand Total

343

The Southern Ute Indian Tribe reported the most point source data to the 2017 NEI with 293 facilities.
This tribe responded to the questionnaire, and the results of this report reflect that response.

It is expected that the CAERS could facilitate emission inventory development by tribes by providing
tribes a way to collect point source data, because in most cases, there are similarities between tribal
emission inventories and State and Local (SLs) emission inventories. By enhancing how CAERS handles
facility inventory data, tribes might find the use of CAERS more advantageous and could be encouraged
to report more so than they do now.

3.4 Facility Inventory Information in Emission Inventories

Twenty-eight SLTs do not collect emission inventory (El) information beyond what is identified in the
CERS 2.0, while 26 SLTs collect El information beyond that found in CERS 2.0. A variety of additional
information collected depending on special SLT needs. Examples are:

-------
• Idaho collects extra geographic data for the facility and release points, design capacity for all
emission units (versus the subset required to do so for NEI purposes), some extra control fields,
and a Federal Employer ID to support emission fee billing.

• Massachusetts collects information on potential-to-emit and permitted limits for throughput
and emissions, detailed information on emission units (especially emergency generators),
monitors, and controls.

• Minnesota collects information on permit type, source type, comment fields for each data
component, and NAICS for emission units, whereas the federal program requires only facility
level NAICS. 'Also, the CERS requires information on one field for the percent reduction achieved
for a pollutant when all control measures are operating as designed for a site control measure.
Minnesota collects information on two fields for a pollutant controlled by a site control
measure: control efficiency and capture efficiency.

• Wyoming collects permit limits at the unit-process level, U.S. Well Numbers for wellhead
facilities, and an Oil & Gas Facility Category for facilities in the oil and gas sector.

• Illinois collects additional information that pertains to air emissions rules specific to the state.

• Colorado collects data related to permit processing and analysis.

• Wisconsin collects number of employees, area from fence line to fence line, whether an
Environmental Management System (EMS) exists and if the EMS is reviewed by a third party,
and whether the facility is a small business (less than 100 employee and annual receipts not
more than $750,000).

Several respondents also indicated that the information they collect for a facility inventory depends on
the type of a facility. For example, North Carolina has facility data for both permitted and permit-
exempt facilities, but only permitted facilities report emissions.

Some SLs also indicated more pollutants were covered in their Els than in the NEI, such as Colorado;
Maine; City of Louisville, Kentucky; Washington; and Wyoming, although pollutant emissions are not
included in the facility inventory data. Therefore, the analysis of pollutant differences is excluded from
this report.

CAERS workflows with SLTs would have to be designed with these SLT-specific data fields in mind. This
means CAERS would have to accommodate additional codes to existing fields, additional data fields that
SLTs require, as well as QA checks, for example¦, to enforce reporting of required fields where the EPA
does not require them.

3.5 Techniques to Obtain and Update Facility Inventory Data in Els

The techniques used to obtain and update facility data existing within Els vary by jurisdiction.
Within any one jurisdiction, they may also be dependent on facility characteristics, such as new or
existing facilities; permitted or not permitted; and permit type. One jurisdiction may use more than one
technique to obtain facility data for emission inventories. For example:

• Oklahoma maintains an Air Quality facility database (called TEAM) used for their permitting,
compliance and enforcement, and emissions inventory programs. Oklahoma collects facility
emissions data using a version of SLEIS from Windsor Solutions. SLEIS and TEAM are not
dynamically connected but data do flow in both directions between the systems in an ad hoc

-------
way. Data in TEAM relevant for Els are only for facility level data, not sub-facility level data.
Facility reporters are responsible for reporting all data for emission units, release points,
processes, etc., that is, data considered to be facility data only within the context of El reporting.
Oklahoma El staff, within operational bounds, also make updates to data during the QC process.
• Contrasting with the above, El Staff at SLTs that have a small number of facilities may only
create and update facility inventory data manually, for example: Hawaii, Lane County, Oregon,
and Chattanooga City, TN.

About 30 (56%) SLT El programs obtain facility inventory data "with data flows", which means they do so
directly from other within-SLT program systems and databases (e.g., permitting, compliance, state
master database), using: system integration with other within-SLT systems, linking with other SLT
databases, sharing tables, and/or snapshot synchronization from other SLT programs and/or databases.
Table 4 shows the number of SLTs that use each technique noting that some SLTs may use more than
one "with data flows" technique.

Table 4. Number of SLTs Using Techniques to Obtain and Update Facility Inventory Data in Els

Technique to Obtain and Update Facility Inventory Data in Els* Number of SLTs

El system is part of an integrated/within state database used by other 19
programs (e.g., permitting, compliance, state master DB)

El system is a unique application/database linked to other state database(s) 12

El system shares certain tables with state database(s) for other programs 8

Snapshot sync** with state database for other programs 3

Manually created and updated by El staff 32

Reported by facility operator 34

*Note: Georgia relies on CAERS for facility inventory information, which was not listed as a technique. Before
using CAERS in 2019 their El system was a unique application/database linked to other state database(s).

**Note: The sync function can be performed automatically or manually, and timing of the sync depends on needs.
Here the emphasis is "snapshot" not live link. For example, Minnesota can sync all facilities or individual facilities
from a state master database automatically at any desired time.

Conversely, it is notable that 24 SLTs do not have data flows such as those described above. Table 5
shows the distribution of the 24 jurisdictions "without data flows". This means El staff and/or facility
operators manually create and update facility inventory data to report relevant facility configuration
changes.

Table 5. Number of SLTs that Create and Update Facility Inventory Data Manually

Jurisdiction

No Data Flow

Total Responded to the
Questionnaire

% No Data Flow from Total
Number of Respondents

Local

75%

State

30%

Tribal

100%

-------
Table 5 illustrates higher percentages of local and tribal jurisdictions without data flows as compared to
states. This result is consistent with the smaller number of facilities needing to be included in emissions
inventories for local and tribal jurisdictions. For example:

• Chattanooga City, TN obtains information for 170 facility sites from permitting staff during
annual inspections. Data are manually entered.

• Clark County, Nevada obtains the facility site information during the permitting process and
then retains it in the El database and reports 70 facilities to NEI.

• At Southwest Clean Air Agency, Washington, facility site data are entered into the El system at
the time of permitting or in response to an enforcement action. The El of Southwest Clean Air
Agency only contains data for 19 facilities.

Table 6 shows the cases where one jurisdiction uses more than one technique to obtain facility data for
facility inventories. Note that values in Table 6 include automated and manually conducted techniques.
For example, the manual (without data flow) Tribal case in Table 5 clearly appears in Table 6; because
the tribal respondent indicated they do manual updates, as well as obtain data from facility operators,
both of which are manual.

Table 6. Number of Techniques Used by SLTs to Obtain Facility Inventory Data

Number of Techniques
Used

Local

State

Tribal

Total SLTs

Total

Although 22 SLTs only use one technique in creating and updating facility inventory data, Connecticut
uses 5 techniques except the snapshot sync with state database for other programs to obtain a
complete set of facility inventory information due to the specific structure of their database and El
system.

The nature of these updates indicates that SLTs with more than one method might need more time to
update their facility inventory than other SLTs before reporting begins in CAERS every year. Therefore,
the time that building and updating the facility inventory takes, should be factored in the annual
workflow process between CAERS and the SLTs in question. Similarly, if the SLTs would like to get updates
on the facility inventory from CAERS itself time would have to be factored into the process so that SLTs
have time for their relevant systems to absorb the changes. Finally, while in the case of El reporting for
the NEI the SLT manages the facility inventory data, it is industry reporters themselves who edit and
report facility data directly to other EPA programs (for example, CEDRI). SLTs who prefer their own
facility inventory databases, rather than direct industry input, would need to work with EPA to design
their workflow in CAERS, so that they may provide input on the correct facility data inventory to be
shared between CAERS and the other federal programs.

-------
3.6 Information Sources for Facility Site, Emissions Unit, and Process

In survey responses, SLTs commonly indicated using multiple information sources to compile their
facility inventory. Some information sources may be popular for one data component while not for
others. Table 7 shows the number of SLTs using the different information sources for facility site,
emission units, and processes. Responses do not add up to the total number of respondents because
multiple sources can be used by an SLT.

Table 7. Number of SLTs Using Different Sources of Information for Facility Site, Emission Unit, and Process

Information Sources

Facility Site

Emission Unit

Process

State permitting program

State compliance program

State master database

Combination SLT permitting
program, compliance program,
and/or master database

Facility operator

NEI facility inventory

Previous El

Auto created by El system

Manually created by El staff

Note: "NA" means the choice was not in the survey questionnaire where only the most common information

sources for the components were listed.

Depending on the SLT, multiple programs can be supported by a central database, which supports
interrelated handling of facility data. The state master database may include databases for a permitting
program and/or a compliance program in many SLTs. Some respondents selected all three sources (State
permitting program, State compliance program, and State master database) when they were asked what
sources they used for creating and updating facility inventory data, while others selected only having a
single master database. The entry in Table 7 listed as "Combination SLT permitting program, compliance
program, and/or master program database," avoids double counting with other rows in the table. For
example, Alaska responded obtaining data from a combination of state permitting program, the state
compliance program, and the state master databases. Alaska is counted as 1 state in the row of "SLT
permitting program, compliance program, and/or master database".

About 74% (40 out of 54) of SLTs are using information directly from combined SLT databases for facility
sites. Although the number of SLTs using the combined information source decreases for emission units
and decreases further for processes, about 61% (33 out of 54) of SLTs are using the combined
information source for processes. This is because SLT permitting, compliance, and/or master databases
contain facility site information but less unit (or process) level information. For example, Oklahoma
maintains an air quality facility database that only contains data at the facility site level, not at the sub-
facility level.

About 80% of SLTs also obtain information on the facility site level from facility operator reporting. The
number of SLTs obtaining information from industry reporting increases for emission units (83%) and
increases further (85%) for processes.

-------
Only few SLTs use information from the NEI facility inventory to initiate some facility and emission unit
data update. More than 57% of SLTs carry the previous El as the starting point for the current El. For
example:

• Washington, Maricopa County, and Arizona use the previous El report as the starting point for a
new report and require facilities to review and update all information.

• Minnesota copies the previous El to the current El, then, only synchronizes new information in
the master database with the current El.

As described at the end of the previous section, for SLTs wanting to use their previous year reported
facility inventory as a starting point, their workflow would have to include steps to ensure data to CAERS
come from SLTs with updated information necessary from all their systems or databases. Also, their
workflow would have to include steps to ensure data from CAERS, once received by the SLT, would be
distributed to any systems or databases the SLT would want to update.

While a jurisdiction may use multiple information sources for facility sites, emission units, and
processes, the information sources used could be dependent on the facility permitting types. For
example, MN obtains information from its master database for permitted facilities that are required to
report to the state El program. However, the master database contains information at different levels of
detail for facilities with different types of permits due to the state El rule requirements. For small
facilities with state general permits, the master database only has information for facility-level
attributes. For large facilities with Title V and state permits other than general permits, available
information includes facility sites and emission units but not processes. For non-permitted facilities that
do not report to the MN El program, such as TRI sources, landfills, POTWs, dry cleaners, and human
crematories, MN uses the Toxics Release Inventory (TRI) and other information sources. In all cases, MN
uses the previous El as the starting point. For a large facility, the facility needs to add new processes or
shut down the processes. For small facilities, the system auto-generates emission units and processes
based on the process source classification codes (SCCs) that are selected by facilities.

These examples indicates that in some cases, it might be beneficial for CAERS to obtain the updated
facility inventory data from the SLT permitting system or database directly, instead of from the SLT El
system that is updated from the SLT permitting system or database. The detailed SLT dataflows should
be considered in the development of the SLT CAERS modules. Where SLTs may not need sub-facility data
for some of their facility types we require additional thinking as to how the CAERS module for each could
accommodate the different types of facilities and thus, potentially different levels of detail in their
inventory.

3.7 Information Sources for Site Control and Control Path

EIS contains information about the types of emissions controls that are present at the facility. Site
control paths provide information on how the site controls relate to each other, and between processes,
and release points at the facility. Recently, the EPA changed the way it collects data on control devices
to improve the configuration of control characterizations in the NEI. CERS vl.O was designed to collect
that data using a control "approach". However, the new way of reporting controls is by including
information on "paths", and this change is reflected in CERS v2.0. While SLTs may have implemented
their El programs based on the AERR with El systems based on the CERS vl.O, those SLT El programs may

-------
not collect the information for control path configurations and need time to catch up to the new way of
representing control information.

Furthermore, while all SLTs are required to submit data for facility site, emission unit, and process to
NEI. However, SLTs are not currently required to collect or submit data for site control and site control
path because the requirements are not included in the AERR. They are also not included as an
automated quality check by EIS or listed in the NEI submission guidance "cheat sheet" for facility
inventory data. The EPA is considering whether new control information included in CERS v2.0 should be
more explicitly required.

Therefore, additional questions were included to understand the variation of SLT data practices with
respect to controls data. Table 8 shows responses about control and control path reporting practices.

Table 8. Number of SLTs Using Site Control Device and Path Reporting Practices

SLT Practice

Site Control Data
Total SLTs

Site Control Path
Total SLTs

Collect and Report

Collect Only

Report Data Only

Neither

Blank

Total

It is interesting to note the number of SLTs that collect and report data to NEI for controls is larger than
those collecting path information. Also, some SLTs may be collecting data but not report it to NEI.
Furthermore, 3 SLTs do not collect data but report data for the control devices and control path
component that they have created. Eight SLTs neither collect nor report control information, and 24
SLTs do not collect or report control path information at all.

Table 9 shows the information sources for control and control paths, including the same 'SLT permitting
program, compliance program, and/or master database' information source as described in Section 3.4.

Table 9. Number of SLTs Using Different Sources of Information for Control Devices and Paths

Information Sources

Number of SLTs
Using Source
for

Control Data

Surveyed*

Number of SLTs
Using Source
for

Control Path
Data

Surveyed**

State permitting program

63%

55%

State compliance program

28%

16%

State master database

17%

13%

-------
Information Sources

Number of SLTs
Using Source
for

Control Data

Surveyed*

Number of SLTs
Using Source
for

Control Path
Data

Surveyed**

SLT permitting, compliance,
and/or master database

67%

65%

Facility reporting

91%

94%

NEI Facility inventory

Previous El

50%

55%

Based on emission units/SCCs

24%

23%

* Forty-six SLTs collect data for site control component.
** Thirty-one SLTs collect data for site control path component.

A high percentage of SLTs, about 67% (36 out of 54), incorporate information from SLT permitting,
compliance, and/or master databases. About 91% (49 out of 54) of SLTs also use information from
facility reporting. About a half of SLTs use previous El reported data as the starting point of the facility
inventory for the next El report. A few SLTs use information from NEI facility inventory. About 24% of
SLTs generate information based on emission units and processes (SCCs).

There are many different requirements and business practices for handling controls data. SLTs collect
data based on SLTs regulations and needs that may not match those of the federal program. To meet
the CERS v2.0 specifications for items required by the AERR, some SLTs may, thus, need to transform
their data either during facility reporting or afterwards when the SLTs elect to report control
components to the NEI. SLTs may make assumptions or gap-fill information with their best knowledge to
transform the data. The following list provides some examples to illustrate the range of SLT practices in
data collection and transformation.

• In ID, controls are reported by facilities. The names for the required control paths needed for
the CERS v2.0 control tables are generated by the Department of Environmental Quality after
receiving data from the facilities. Facilities do not see them. Controls are then manually added
to the CERS staging tables. ID reported control path names to EPA for the first time for the 2020
NEI.

• In OK, facilities report control device sequence, percent capture, percent uptime, controlled
pollutants by the control device, and control percent reduction efficiency. These data are carried
forward into the next inventory cycle and can be amended by reporters as equipment
configurations and/or processes change. OK has not implemented the site control path as a
reporting requirement or as an option in the front end of its SLEIS (OK's custom reporting
system). OK transforms its data into the new control path schema during the EIS submission
and aligns with CERS v2.0.

• MA has built simple linear path data into their CAP emissions reporting system. More complex
paths are rarely observed at facilities in MA, and MA is considering the best approach for future
implementation.

-------
• The CT emission reporting system has a sequence order included in the reporting of controls but
is based on a simple unit-level control that does not model controls at a facility level, or
sophisticated control configurations, such as controls in parallel.

• In Michigan, emission units are associated or linked with controls in the state system. Facilities
can select which controls are installed on each emissions unit when filling out their annual
emissions report. Additionally, Ml creates an Excel template to collect supplemental control
information from facilities to populate the CERS v2.0 data fields. This Excel template supports
facilities creating control paths to the associated processes (SCCs).

• Puerto Rico does not collect control path data but creates control paths using the control data
available from facilities and reports it to the NEI.

• WA manually creates new tables using previous data to meet AERR reporting requirements, and
CERS v2.0 format definitions. WA does not collect "Sequence Number", so the state artificially
creates those automatically in its software to convert its data into the XML format for EIS.

• The MN El system is designed based CERS vl.O. It only collects control information for large
facilities from its master database where site controls are connected to emission units. One
emission unit can connect to multiple site controls either in parallel or in series, but there is no
information on connections between site controls. The code that exports data from this
database to be submitted to the NEI has built-in assumptions to allow upload using CERS v2.0.
MN does not collect and report control information for small facilities and non-permitted
facilities.

CAERS reporters are currently submitting data to EIS directly and data are formatted in accordance with
the CERS v2.0 specification. The CAERS schema allows facilities to report data under both federal
requirements by the AERR and SLT-specific requests. The SLTs can request data fields in CAERS as a part
of their overall SLT inventory. While CAERS may be able to collect SLT-specific data, special care must be
taken to ensure that technological artifacts (artificial requirements) are not produced when designing
technical solutions, and do not create unnecessary difficulty for industry reporters or SLTs.

3.8 Information Sources for Release Point and Release Point Apportionment

Although SLTs must identify release points and release point apportionments when submitting
data to the NEI, SLTs have diverse requirements, business practices, and approaches in data handling.
There are two types of release points: stacks and fugitives. Table 9 shows the results of the survey
question asking about the data source for release point apportionment data.

Table 10. Number of SLTs Using Different Sources of Information for Release Points and Release Point Apportionment

Information Sources Stack Release Fugitive Release Release Point

Point

Apportionment

23
11
7

State permitting program
State compliance program
State master database
SLT permitting program, compliance
program, and/or master database

Facility reporting
NEI Facility inventory

45
4

37
17
10

29
13
8

38
6

37
3

-------
Information Sources

Stack Release
Point

Fugitive Release
Point

Release Point
Apportionment

Previous El

28

23

20

Auto created by El system

9

10

8

Manually created by El staff

18

16

20

The observations again show a strong dependence (at least 50%, 27 out of 54 SLTs) on SLT permitting
program, compliance program, and/or master database as sources of information for release point and
release point apportionments data. This reliance on SLT non-El programs and databases for release
points and release point apportionment information is less than for data on facilities, emission units,
and processes (at least 61%, 33 out of 54 SLTs). This result shows that SLTs are less likely to have
information readily available for release points and release point apportionments than they are to have
facility-, unit-, or process- level data. Therefore, they may be having to take additional steps to create
and report release point data to NEI.

Figure 3 illustrates the reliance of data sourced from non-El datasets for facility and sub-facility
information. It compares each of the data types previously described by showing the percentage of SLTs
that rely on other databases within the SLT to get that information.

Figure 3. Percent of SLTs Relying on SLT Permit Program, Compliance Program and/or Master Database for Facility and Sub-
facility Data

Facility Site Emission Unit Process Stack Release Fugitive Release Point

Point Release Point Apportionment

Because of the lack of release point information, it is possible SLTs will rely on their NEI facility inventory
data more or exclusively to maintain and update their release point information. Specifically, 4 SLTs use
NEI facility inventory as the source for their stack parameters and 6 SLTs use it for fugitive release
parameters. For the release point apportionment, 3 SLTs use the NEI facility inventory, which is more
than those that use it for facility (2 SLTs) and emission units (1 SLT). Fewer SLTs use previous El
information stored at the SLT, particularly for release point apportionments (36%), than for facility sites,
emission units, and processes (for which at least 53% of SLTs use the previous El). Many SLTs need their

14

-------
El information from the NEI on release point apportionments and paths, as that information is recently
required and SLTs may not have had time to adjust their other facility inventory sources to intake and
update the new information if they desire to have it stored there.

Facility reporting is the most popular information source of release point and release point
apportionment information. Finally, about 15% of SLTs use information created manually by El staff and
10% use data created automatically by the El system.

A single jurisdiction may use a variety of approaches to obtain data on release points and release point
apportionments. The data collected from facilities may be different from the data a SLT transforms and
submits to the EIS. SLTs must use assumptions and gap-filling in data transformation to the CERS v2.0
format. In doing this, some SLTs collect total emissions and assume 100% of emissions go through
stacks, and don't consider fugitive release points at all, even though doing so is not expected by the
AERR. Examples are provided below.

• MA does not collect release point apportionment data. The release point apportionment
information is auto created by their El system based on some very basic assumptions.

• Pima County, AZ, has not collected or reported release point data for fugitive emissions.

Fugitive emissions have been reported but not the associated release point data. Therefore, all
emissions (100%), including fugitive emissions, are reported as being released through stacks.
Because Pima is a CAERS user, and because CAERS is set up for facility reporters to apportion
data to fugitives, in future Pima could refine its data on fugitive emissions.

• In Wl, the El system assumes all emissions from emissions units are fugitive if the emissions are
not partitioned to any stacks.

• Indiana's database does not support apportionment. IN gathers the information through an
online system used to prepare the emissions statements. IN saves the data in an Oracle
database based on a very early version of the NEI, then, submits it through the node. IN
assumes 100% of emissions are released through stacks.

• Ml created an Excel template in addition to its El system to fulfill the requirements of the new
CERS v2.0. This includes the creation of release point apportionments for any associated stacks.

• MN's master database contains information for about 428 large facilities on stacks but only
identifies fugitive emission units, not fugitive release points. Fugitive release points are auto
generated by the El system. MN does not report any parameters associated with fugitive release
points, such as width and length. For about 1,700 small permitted and non-permitted facilities,
the El system auto generates one dummy release point (could be a stack or a fugitive release
point) for an emission unit based on the first process SCCs. The dummy release point is used by
all processes under that emission unit.

• Lincoln/Lancaster County, Nebraska assumes 100% emissions are released through stacks.

• In CO, process-pollutant level emissions are 100% assigned to a single stack (AIRS/AFS model).

• HI uses the default SLEIS input format for 2020 emissions. This means collecting data meeting
the CERS 1.0 format and then letting the Bridge Tool convert to the CERS 2.0.

Because CAERS is already set up to obtain detailed data for release points, SLTs taking advantage of
CAERS for reporting would benefit by getting this data from industry reporters. In addition, having
data for release point apportionments include fugitives would mean a better alignment with TRI
reported emissions, as TRI requires fugitives to be reportedand there can be a mismatch in
reporting for both NEI and TRI solely due to the fact that a facility may be reporting fugitives to TRI
but not NEI.

-------
4 Discussion and Recommendations for CAERS development

The Facility PDT R&D team discussed the survey responses and their implications for CAERS. The
following recommendations were issued from those discussions.

1. CAERS capacity: CAERS needs to be able to handle a large volume of facilities, about 30,000
facilities in one jurisdiction. This need applies to data management within CAERS and data
transfer between CAERS and SLTs. Additionally, many SLTs contain non-NEI reporting facilities in
their inventories and some SLTs keep such facilities in EIS already. CAERs would be used by SLTs
not only for facilities that are subject to federal-reporting requirements but also for non-federal
reporting requirements in use cases 2-4 of SLT workflows. Otherwise, the SLTs would need to
have one reporting system for facilities that also report to one or more federal programs, and a
separate reporting system for those that don't. CAERS system performance should not be
slowed down for SLTs with a large volume of data.

2. SLT-specific data fields beyond NEI requirements: These include:

• data fields,

• codes, and

• types of facilities including differing levels of detail for facility and sub-facility level data for
the facilities depending on their type (for example, size, type of permit). Some implications
of these differences are:

a. CAERS workflows with SLTs would have to be designed with SLT-specific data fields in
mind. This means CAERS would have to accommodate additional codes to existing
fields, additional data fields that SLTs require, as well as QA checks, for example, to
enforce reporting of required fields where the EPA does not require them.
Considerations include.

b. The facility inventory information beyond CAERS is not needed by more than 50% of
SLTs. The CAERS development team should use the information collected here to help
decide which data elements should be included in the CAERS core information model,
which should not be included, and which could optionally be included depending on
further evaluation and specific SLT needs.

c. The need for CAERS to support facility inventory information beyond CERS v2.0 depends
on SLTs regulations and requirements. CAERS will support SLT-specific facility data fields
already included in CERS v2.0 when webservices between CAERS and SLTs are
established. In addition, at the time this report was being finalized, the PDT had
discussed additional SLT-specific data fields that SLTs may want to see in CAERS in
future. The need for one jurisdiction may not be a need for others, and CAERS should
account for that.

d. The CAERS codes and functions (including QA checks) for using those data fields that
meet SLT-specific needs should be considered in future CAERS versions as SLTs onboard
so their specific needs may be met. However, where an onboarding SLT would like to
take advantage of the fact that CAERS is already set up to take information in using a
certain format, the SLT may want its module to take advantage of this set up. For
example, through CAERS, the facility reporter may add fugitive release points, and
associate controls and control paths to stack release points. This data could also be fed
back to the SLT.

-------
3. Differences in SLT data collection and updating methods: Many SLTs, from 31% to 59% of

respondents depending on the components, use their previous El as a starting point for the
current El. CAERS already copies the previous year El to the current El as a starting point for
facility reports as a default. However, many SLT's El systems have inherent relationship with
other SLT's programs such as permitting and compliance. The systems for those other programs
are also SLT-specific. CAERS could interreact with SLT emission inventory systems and/or directly
interact with databases for other SLT programs (for example, permitting) depending on the SLT's
specific needs. Or the SLT could create, as part of its workflow with CAERS, the compilation of
the facility inventory data that the SLT wants to have in CAERS for reporting; SLTs could develop
more specialized modules with guidance and help from EPA for this purpose. The SLT-
specialized CAERS module could have specialized interfaces and perhaps special web services
that use the SLT data fields between the CAERS facility data and the multiple SLT data systems.

There are multiple scenarios that CAERS should handle:

a. Differences in the SLT starting facility inventory for a given reporting year.

i. An SLT may rely on CAERS exclusively to obtain facility inventory updates as
reported by industry.

ii. An SLT may rely on some of the data in CAERS but may want to supplement the
facility inventory with updates from some of its other databases or systems,
and/or manually.

In either of these cases, if the SLT would like to get updates on the facility
inventory from CAERS, this would require a workflow that would allow them to
get that data when and how they'd like to get it. In addition, if the SLT would like
to parse out the facility inventory data to update different SLT databases and
systems this step would also have to be factored into its overall workflow with
CAERS.

iii. The SLT may not want to rely on the data in CAERS as a starting point for the
facility inventory to start reporting for a given inventory year and thus, may
want to pull its facility inventory into CAERS every year.

SLTs that would like to have their El in CAERS before annual reporting begins, and
who use multiple sources of data would need to factor in more time to update
their facility inventory in CAERs and their workflow with CAERS could involve
coordination amongst multiple databases. If those SLTs would also want to take
advantage of CAERS to allow facility reporters to update the facility inventory,
they could do so with the help of QA checks in CAERS that would ensure data is
entered correctly. Then, a workflow that brings those updates back to the SLT
database(s) would have to be designed and implemented.

SLTs who prefer their own databases for facility inventory updates, rather than
direct industry input, would also need to think about when and how to update
CAERS with the goal of ensuring their most updated facility inventory is shared
with other EPA programs such as CEDRI, for example. This is because for other
EPA programs, industry reporters directly report and edit the facility data
themselves, and thus, the SLT would have to weigh in to determine the correct
version of the facility's inventory data to be shared from CAERS. Further research
and discussions are needed to this effect.

-------
4. Control device data: In terms of controls and control paths, there are three broad use cases for
SLTs:

a. SLTs require detailed control and path information: CAERS is already equipped to allow
facility reporters to build paths from processes to release points and "reuse" those
paths where applicable to more process-release point combinations. While the process
for collection control data is straightforward for facilities with single controls between
processes and release points, complex control configuration sometimes require
assistance from SLTs and EPA to set up appropriately in CAERS. And this assistance has
been available to facility reporters since CAERS MVP. However, in the dataflow between
CAERS and the SLT, the SLT may require additional functionality to be able to update the
information in CAERS as needed, if it's facility inventory will not be based on the data in
CAERS.

b. SLTs do not have rules to require facilities to report but encourage facilities to report:

Ideally, CAERS could include a mechanism to auto-generates control path configurations
for SLTs that do not require that information. While the data would not be as robust,
this could be done based on existing information for controls, their connections and/or
SCCs. The need for this feature in CAERS is highly dependent on the need to support
different SLT regulations and practices, which could change over time depending on any
future changes to the AERR. Because CAERs already has the capability to obtain control
path data from facility reporters, SLTs may want to take advantage of CAERS reporting.
Where SLTs wish to send their data to CAERS (Case 1 SLT) their use cases would have to
be reviewed to see how this automated functionality might be added to their module.
This means that the workflow between CAERS and SLT would have to include additional
code for this transformation.

c. SLTs do not have rules to require facilities to report and do not see the value of
collecting detailed information on site control component and site control path
component in the SLT: The workflow would not require CAERS to return the updated
control and path data to the SLT. In this case, the SLT-specific CAERS module could auto
populate CAERS site control component and site control path component with the
control information at a higher level (such as 80% of control for PM10-FIL) and basic
assumption provided by the SLT.

5. Release point data: Handling fugitive release points and release point apportionment is
another challenging task. Some SLTs may not have regulations to support it and facility reporters
may not report what is needed in a correct way. Although some SLTs allow facilities to report
the information, facilities may not do a diligent work to obtain the information without SLT
requirements. Assumptions have been made by SLTs to be able to include this information in
the NEI submissions. SLTs adopting CAERS, may benefit from allowing their facilities to report
the release point apportionments, and incorporating that information into their systems, or
apply assumptions to their data in CAERS so it is set up with fugitive release information before
sending it to EIS.

The same applies to release point parameters, such as stack heights, stack diameters. Currently,
some SLT El systems auto fill the missing values by using SCC-based default stack parameters. If
those kinds of needs arise, CAERS or most likely, SLT-specific CAERS modules should have this
function. The current CAERS has QA checks to make sure no data are missing.

-------
6. Time series: While not explicitly mentioned in the survey analysis, the team discussed that it is
part of a good facility inventory to be able to track components through time. Date stamps
would be helpful to include for many facility inventory data fields. This time-stamp-ability is
already being explored in CAERS so that SLT-specific start and end date type fields may be
included in future for SLTs that would like to have them. As new data is added these time
stamps may also be generated in CAERS.

7. Data editors: Another aspect the team derived from the above discussion that it deemed
important, and has already been mentioned, has to do with who should edit which parts of a
facility inventory. SLTs may want to prevent facility reporters from editing certain data fields.
SLTs need to be able to manually create and update certain facility inventory data before
submitting data to NEI by El staff. Different SLTs may require different data field restrictions.
CAERS already allows SLTs to determine which data they enter versus which data they want
facilities to enter. To alleviate burden to SLTs, an option that some onboarded SLTs have opted
for is to issue warnings when a facility reporter has edited certain fields. This allows the SLT to
be aware and accept the change if desired, without having to spend SLT staff time entering the
data themselves.

8. Editing functionality: Given the variety of needs and that some SLTs have certain functionalities
to assist in the creation and editing of their inventories, the team considered that it would be
helpful for CAERS to have future enhancements or new capabilities to allow copy/paste records
and auto fill of sequential numbers to make manual data editing easier and less time consuming.

9. Reporting by tribes: Noteworthy was the fact that only one tribal jurisdiction responded to the
survey. Tribes are not currently compelled to report to NEI, and in may not have the resources
to do so. CAERS could facilitate emission inventory development by tribes by providing tribes a
way to collect point source data, because in most cases, there are similarities between tribal
emission inventories and SLs emission inventories. By enhancing how CAERS handles facility
inventory data, tribes might find the use of CAERS more advantageous and could be encouraged
to report more so than they do now.

10. Future research.

a. Facility management: EPA has the Facility Registry Service (FRS) integrates facility
information from several different national and state information systems
(https://www.epa.gov/frs), including compliance and permit data for stationary sources
of air pollution regulated by the EPA, state, and local air pollution agencies. The R&D
facility sub team explored the possibility to use FRS as one information source for
CAERS. It was found that while initial work had already been conducted to explore the
possibility of using FRS as a data source for facility information in CAERS, including
workflows with the SLTs, FRS is not designed to meet future CAERS needs in terms of
facility inventory data sharing with other federal programs or SLTs without major
overhaul of the system, so at this time FRS is not feasible to be used with CAERS.

b. QA/QC checks: A subsequent report from this team will explore the types of quality
assurance (QA) and quality control (QC) checks that CAERS should have to allow SLTs to
customize their use of CAERs.

-------
c. SLT facility inventory updates and how they should be shared with the other federal
programs via CAERS is also a topic for future research. Especially in situations where
SLTs do not wish to rely on facility reporters to obtain facility inventory information.

Integration between CAERS and SLT systems will be crucial to keep facility inventory data updated. Also,
El systems are linked with other SLT systems. This means that there are different cases for CAERS facility
workflows ranging from SLTs who would use CAERS as their starting point for reporting the following
year, to workflows between CAERS and SLT facility inventory databases, and potentially allowing CAERS
to receive certain data from different SLT databases to mimic that SLTs current practices. Differences in
the facility inventory content (types of facilities, data fields and codes) beyond the NEI also exist. CAERS
will require much flexibility with respect to facility inventories to accommodate future SLTs and their
workflows.

20

-------
Appendix A - Original Letter and Questionnaire

See Appendix A. Original Letter and Questionnaire.docx

Appendix B. EIS facility staging table requirements

See Appendix B. Facility Staging Requirements.xlsx

Appendix C. Responses to the Questionnaire

See Appendix C. All Original Responses.pdf

Appendix D. Data analysis

See Appendix D. SLT Facility Inventory Research.xlsx

21

-------