DSHW-2018-003922 - 0901a0688080c67d

Div of Waste Management
and Radiation Control
Chevron
1%0
IMO
John Amato Project Manager
MAY 03 2018 7-44 V V-ZI
Refining Business Unit
Chevron Environmental Management Company 324 W. El Segundo Blvd. (M0-133) El Segundo, CA 90245
[phone] [e-mail]
April 27, 2018
Mr. Scott Anderson
Division Director
Division of Waste Management and Radiation Control
Utah Department of Environmental Quality
195 North 1950 West
P. O. Box 144880
Salt Lake City, Utah 84114-4880
RE: Response to Review Comments on Contaminant Transport Model Update
Chevron Products Company, Salt Lake Refinery, Salt Lake City, Utah
Dear Mr. Anderson:
On March 12, 2018, Chevron received comments from The Utah Department of Waste Management and Radiation
Control (Department) seeking further clarification on several aspects of the Groundwater Flow and Contaminant
Transport Model Update submitted in December 2017. EarthFax Engineering Group, LLCs (EarthFax) was tasked
with preparing the responses to the Department's review comments. The memorandum from EarthFax containing
both the Departments review comments and their responses is attached.
If you have any questions or comments regarding the attached memorandum, please contact me at [phone] or
via email at John.Amatogchevron.com.
Sincerely,
JohI Amato
Project Manager
Chevron Environmental Management Company
Cc: Hao Zhu, Utah DEQ
Keith Rittle, Trihydro
Christina King, Chevron SLR
TECHNICAL MEMORANDUM
TO: Mr. John Amato, Chevron Environmental Management Company
FROM: Richard White, EarthFax Engineering Group, LLC
DATE: April 26, 2018
RE: Response to Comment on the 2017 Chevron Salt Lake Refmery Groundwater
Flow and Contaminant Transport Model by the Utah Division of Waste
Management and Radiation Control
In December 2017, Chevron Environmental Management Company ("CEMC") submitted a
report to the Utah Division of Waste Management and Radiation Control (the "Division")
entitled Chevron Salt Lake Refinery Facility Groundwater Flow and Contaminant Transport
Model Update, prepared by EarthFax Engineering Group, LLC ("EarthFax"). The purpose of
that report was to update the results of prior modeling of groundwater flow and contaminant
transport performed by EarthFax at the site beginning in 1992 and as subsequently updated in
1996, 2012, and 2014. In accordance with a post-closure permit issued to Chevron originally in
1997, the 2017 model also included a Monte Carlo uncertainty analysis.
On March 12, 2018 the Division issued comments to John Amato of CEMC concerning the 2017
model. The purpose of this document is to respond to the Division's March 2018 comments.
For the sake of completeness, the Division's comments are presented below in bold italics with
responses presented in normal text.
I) Chapter 2, page 3 and 4. The text mentions that an upward gradient exists between the
deeper, confined aquifer and the surface aquifer (the focus of the update). For clariV, please
state how many locations this was observed at, and y' any current data exists to confirm these
observations which date back to the 1992 and 1996 model reports.
Historical groundwater quality data collected from deep monitoring wells at the Salt Lake
Refmery consistently indicated a lack of detectable hydrocarbon concentrations. Therefore, in
accordance with the facility post-closure permit, all wells monitoring the deep, confmed aquifer
were removed from the groundwater monitoring program in 1997. Several of these wells have
since been abandoned in accordance with Utah standards to accommodate infrastructure
modifications at the refmery. As a result, no local data have been collected since 1997 to
confirm the upward groundwater gradient at the Salt Lake Refmery. However, long-term data
downloaded from the U.S. Geological Survey Active Water-Level Networki, as summarized in
Figure 1, indicate that groundwater levels in 2017 collected from wells completed in the regional
confined aquifer near the Salt Lake Refinery are consistent with those collected in the early
1990s when the flow model was first calibrated. Therefore, it is reasonable to conclude that the
upward gradient between the deep, confined aquifer and the shallow, unconfined aquifer
continues to exist at the Salt Lake Refinery.
https://awatenusgs.gov/infodata/groundwater.html
Mr. John Amato Chevron SLR 2017 Groundwater Model
Chevron Environmental Management Company Response to Comments by Utah DWMRC
April 26, 2018
The Classic interface was used during this modeling effort. This interface affects only the screen
presentation, not the operation of the model, which was developed and first published in 1984
under the authorship of M.G. McDonald and A.W. Harbaugh of the U.S. Geological Survey, as
noted in the 2017 model report.
SimLab is a public-domain simulation package used for uncertainty and sensitivity analysis.
This software was developed by the European Commission's Science and Knowledge Service,
Joint Research Center, and is available at https://ec.europa.eu/jrc/en/samo/simlab. As stated in
Section 4.1.1 of the 2017 report, Version 2.2 of Simlab was used for the Monte Carlo analyses.
The organization that controls the development of the software and the web address from which
this software was downloaded were cited in the text. Since neither the web site nor the Simlab
reference manual list authors for the software, this form of citation is considered standard.
However, at the request of the Division, a Simlab citation will be added to the list of references
in future model update reports.
4) Section 3.1.1, Model Setup, page 6. The text states that it was assumed that principal
directions of hydraulic conductivity are collinear with the x and y coordinates of the model
grid. While this was the case for the 1992 and 1996 model setups, how was that verified in
terms of the current model?
As noted in Sections 2.3.1 and 3.2 of the 1992 groundwater model report, the principal direction
of hydraulic conductivity for the shallow aquifer beneath the Salt Lake Refinery, based on the
results of a single pumping test, is offset 9° from a true east-west line. Given this small
variation, prior groundwater modeling efforts were based on a north/south-east/west coordinate
system. Given the fact that there have been no changes to the hydraulic conductivity of the
shallow aquifer since 1992, together with the minor directional variation from standard
coordinates and in order to maintain consistency with past modeling efforts, the 2017 model was
also performed based on a north/south-east/west coordinate system.
5) Section 3.1.1, Model Setup, page 6. The text states that hydraulic transmissivities (and
therefore also conductivities) are based on pumping and slug test data, presented in Appendix
A, but were varied for model input based on professional judgment. Please explain what the
professional judgment consisted of A perusal of Appendix A data gives the impression that
perhaps early slug drawdown responses were matched, rather than employing the suggested
"B-C" fit, as recommended by Bouwer, 1989, The Bouwer and Rice Slug Test - An Update,
Groundwater, Vol. 27, Issue 3. Please elaborate.
Appendix A of the 2017 model report presents the results of slug tests performed on seven
monitoring wells at the Salt Lake Refmery in June 2017. These wells were all installed after the
early 1990s when hydraulic testing was conducted in the then-existing facility monitoring wells
in support of the 1992 groundwater flow model
3
Mr. John Amato Chevron SLR 2017 Groundwater Model
Chevron Environmental Management Company Response to Comments by Utah DWMRC
April 26, 2018
The 1992 groundwater flow model utilized the parameter-optimization program MODINV
(developed by Doherty [1990]) to improve model calibration. As stated in the 1992 model
report, "using MODINV, specific values . . . can be optimized such that model-generated water
levels are as well matched as possible to those observed in the field. . . . The objective of
MODINV is to minimize the sum of the squared residuals (i.e., differences between simulated
and observed water levels). . . . Initially, an automated calibration using MODINV was
performed. This allowed the computer to perform the time-consuming trial-and-error iterations to
approximate the initial water levels. After a satisfactory initial calibration was obtained by using
MODINV, the model was fine-tuned by using manual trial-and-error methods."
The software package PMWIN (Processing Modflow for Windows), developed by Chiang and
Kinzelbach (1996), was used during calibration of the 1996 groundwater flow and contaminant
transport models. This package was used to design the model structure and to create, adjust, and
manipulate input and output files, thereby simplifying model setup and interpretation.
Given the optimization efforts that went into the original setup of the groundwater flow and
contaminant transport models, together with the fact that subsequent updates were conducted
primarily to account for changes in site conditions, additional optimization via add-on programs
was not considered necessary. Since the base model had been optimized previously using
automated methods, manual methods were considered an adequate optimization approach for the
2017 model update.
Other calibration-oriented figures beyond those presented in the 2017 model report were not
prepared since the model objective functions were met and this effort was intended primarily to
update the model to current facility conditions. The use of additional calibration-oriented figure
will be evaluated for use during future model updates.
7) Section 3.2.1, Model Setup, page 8. The text mentions that constant concentration cells
were used to model sources. This seems to be a valid approach if the intent was to be
conservative, but it could be more realistic to model these cells as continuous sources with
decaying concentration (provided calibration is successful). The benefit might be that
predicted, probabilistic contaminant concentrations would be lower in the model runs for the
years 2037 and 2047, as displayed in Appendices C and D (which is what one would expect).
The only sources allowed by Visual Modflow are constant concentration, recharge concentration,
and point source. Contaminated groundwater at the Salt Lake Refinery exists in a relatively
small area and has existed in that small area for several years. Given the length of time that
contaminants have existed in this limited area, it is reasonably likely that these contaminants
exist as residuals that are sorbed to the soil. Thus, of the available model source settings,
constant concentration sources were considered most applicable. However, Visual Modflow
does allow the use of first-order irreversible decay downgradient from a constant concentration
5
Mr. John Amato Chevron SLR 2017 Groundwater Model
Chevron Environmental Management Company Response to Comments by Utah DWMRC
April 26, 2018
10) Section 4.2, Analysis Results, p. 15. The Division recommends adding a short summary
discussing lessons learned and recommendations for future groundwater sampling and for
future groundwater and transport modeling updates.
A brief summary of lessons learned can be added to future reports to discuss recommendations
for future groundwater sampling and modeling. The results of the 2017 modeling effort, together
with the most recent (future) groundwater sampling results, will be used during discussions prior
to the next post-closure permit update to assess the adequacy of the monitoring program (in
terms of the number and locations of the wells being sampled and the parameters for which those
samples are being analyzed).
11) Appendix B, Comparisons of Converged vs. Non-converged Realizations. While it appears
that converged and non-converged distributions are sufficiently similar for the parameters
shown in Appendix B, is the presented chart gallery of parameters exhaustive or selective?
Please elaborate.
The chart gallery of parameters is exhaustive.
12) Appendix C and D, Predictive Probabilities for Benzene and Toluene in Select Monitoring
Wells. It appears that all of the probability vs. concentration charts have gaps with
probabilities of zero at irregular interval s. Is this a printing artifact or a function of the
limited number of runs which converged? Please explain.
As in all sampling efforts (whether physical or statistical samples), the goal of sampling is to
obtain an understanding of the population with a reasonable degree of confidence. The only way
to effectively fill the concentration gaps is to increase the number of Monte Carlo samplings. As
noted in Section 4.1.3 of the 2017 report, this evaluation was conducted with 207 converged
simulations, compared with a minimum desirable sample size of 130 simulations. Therefore,
nearly 60% more converged simulations were evaluated than recommended as a minimum.
Performing more simulations would decrease the number of gaps but would not add materially to
an understanding of the population.
References Cited
Anderson, M.P. and W.W. Woessner. 1992. Applied Groundwater Modeling. Academic Press,
Inc. San Diego, California.
Bouwer, H. 1989. The Bouwer and Rice Slug Test - An Update. Ground Water. Vol. 27, No.
3, pp. 304-309.
7
USGS 404826112062201 (B- 1- 2) 9aca- 1
-9.8
• mmimmmwmaammumm....m.m.414mmumlint.www_ammmintalmo
I- 4228.0
4227.8
I- 4226.0
4225.8
„., 4224.8
1- 4223.0
4222.8
o
721
Gr
o
u
n
dw
a
te
r

le
v
e
l
a
bo
v
e

NG
V
D

1
9
2
9
,

fe
e
t
", 1- 4221.0
1964 1970 1976 1982 1988 1994 2800 2886 2812 2018
-17.8
-16.0
-15.0
Period of approved data Period of provisional data
USGS 404 6 1523301 D- 1 ) 7abd- 6
... Period of approved data mmm Period of provisional data
4275
4270
4265
4260
4255
4250 Gr
o
u
n
dw
a
te
r

le
v
e
l
a
bo
v
e

N
G
V
D 19
2
9
,

fe
et
c -15
3 O -10
t. 4.) 3
O ea .c 4.) a. 10
1940 1952 1964 1976 1988 2808 2812
A) Well located approximately 8.5 miles WSW of the Salt Lake Refinery
B) Well located approximately 5.5 miles SE of the Salt Lake Refinery
Graphs downloaded from httos://utwater.usgs.gov/infodata/eroundwater.html
FIGURE 1. HISTORIC GROUNDWATER-LEVEL DATA COLLECTED FROM
THE REGIONAL CONFINED AQUIFER
=AOC
-4AZ41 IA, lel • V18mdrA
EarthFax