June
2016
HYDROCARBON
ENGINEERING
42
n
n
Determining precision:
§
§
1.5 times the repeatability ‘r’ divided by 2.77, where ‘r’
equals the ASTM repeatability of ASTM D7039.
For example, considering a 10 ppm sulfur gasoline sample,
the maximum allowable standard deviation of 20 tests is
≤1.5 * (1.73 ppm/2.77) = 0.94 ppm.
The following exemplifies the calculation of performance
accuracy according to ASTMD6708 and performance precision:
n
n
Routine performance accuracy:
§
§
Initial work: test a certified gravimetric standard (CRV) a
minimum of 15 times to acquire setup data.
§
§
Refer to ASTM D6299 Section 8.2 ‘Case 1’. Pre-treat data,
construct a run chart, and confirm (|Mean-CRV|≤0.75r –
according to industry sources, the intent of this section
of the regulations is unclear regarding confirmation.
Typically, the EPA will publish a question and answer
section to clarify, if deemed necessary).
n
n
Construct (I) and (MR) charts:
§
§
Ongoing work: maintain routine testing of CRV, minimum
of three per year.
§
§
Chart and monitor the moving range and all individual
data.
§
§
Apply run rules and specifications for the data, and
perform ACCA on violations and outliers.
n
n
Performance precision:
§
§
Test a qualifying QC material either once a week, or
every 20 production tests, whichever is more frequent,
and construct (I) and (MR) charts.
§
§
Validate new QC either by parallel runs or Q tests.
§
§
Document, file and retain method qualification data as
per the legal department. Note: the EPA requires a
minimum of five years. Formal assignable ACCA
documentation is required for all SPC data.
Statistical software solutions to help
with Tier III compliance
Developing data analysis solutions for the PBATMA involves the
application of SPC, such as controlling the process of running
tests for the certification of fuel. Fortunately, there are software
solutions available to help refineries comply.
n
n
The EPA offers an online spreadsheet to log, verify and
submit precision and accuracy on diesel sulfur. While this
spreadsheet is designed for diesel sulfur, it can be readily
modified for use with other parameters. Figure 3 shows a
sample page of the EPA sulfur test method spreadsheet.
§
§
SQC/SPC software is an integral part of compliance with
the EPA PBATMA.
n
n
A reliable and easily installed system provides a clear
display of compliance with the regulations. Some of the
key attributes include:
§
§
The ability to configure charting details to match
compliance criteria such as run rules (ASTM, not
Western Electric), control limits and ACCA display
call outs (Figure 4).
§
§
The ability to be easily understood and modified by
onsite staff, not IT.
n
n
Manufacturing intelligence solutions (e.g., NWA Focus EMI®
from Northwest Analytics) are used by the chemical,
petrochemical, speciality chemical and materials
manufacturing industries to generate analytics-based
alarms in real time for immediate issue identification
before final product quality is impacted. This has paved the
way for meeting current EPA requirements since these
systems promptly note the deviation of parameters from
expected trends. A clear environmental dashboard, as seen
in Figure 5, provides constant, real time tracking.
n
n
Key attributes of a well designed MI system include:
§
§
The ability to access data directly across multiple,
disparate data sources (e.g., LIMS, MES, process
historians).
§
§
Off the shelf configurability (no customisation
requirements).
§
§
Global analytics layer.
§
§
Easily understood analytics; without the need for data
scientists (no ‘black box’ analytics).
§
§
Role specific visualisation/dashboards.
§
§
Enabling changes by users without the need for
IT intervention or involvement.
§
§
No data duplication (the use of existing process
databases).
Trusting the numbers
As previously mentioned, the numbers produced in the fuel
refining industry must be accurate and verifiable. The precision
and accuracy of data are critical for Tier III compliance as well
as for business and operating decisions. However, many
‘canned’ systems and customised in-house solutions replicate
databases rather than working with existing databases, which
impacts the reliability of the data collected. The only way to
produce truly reliable data is to use existing databases in
real time for statistical analysis. Figures 6 shows software
tracking statistical trending and the provision of alarms for
laboratory personnel.
Conclusion
Dependable statistical software must use verifiable statistical
methods so that continuous and real time monitoring will result
in reliable action items and regulatory information. By detecting
precise events, as well as recognising trends, the analytical
system can track all aspects of the process and display the
status of every parameter.
One glance can reveal a wide range of information that can
be quickly understood and managed. Easy auditing in real time
can be achieved by using a flexible dashboard display that
organises necessary information for easy access for different
access levels throughout the company.
Notes
This article was written by John Maurer, JSmartpm, LLC, USA, on
behalf of Northwest Analytics.
References
1.
LIN, E. W., MUKHERJEE, U. J., and RISSE, P. J., 'Clean by Design',
Hydrocarbon Engineering, March 2014., retrieved from
com/images/uploads/technical_articles/Clean_by_Design.pdf.
2.
Current code of federal regulations related to EPA Tier III performance-
based methods – Title 40 located in Volume 17, Part 80, Subpart D,
Section 47:
/
Title40/40tab_02.tpl.
3.
The following link points to the subject matter involving Tier III
performance based methods and Title 40, located in Volume 17,
Part 80, Subpart D, Section 47:
-
idx?tpl=/ecfrbrowse/Title40/40tab_02.tpl.
4.
As explained in OMB Circular A-119 and the National Technology
Transfer and Advancement Act of 1995, P.L. 104 - 113, section 12(d).
5.
EPA spreadsheet to log, verify and submit precision and accuracy on
diesel sulfur:
-
compliance-help/sulfur-test-facilities.
