Dixie Fire Investigation Report
Dixie Fire Investigation Report
Date and Time of the Incident: July 13, 2021 at 0648 hours
Fatality/Injury: 4 injuries
Violation: Yes
I. Summary
On July 13, 2021, at 0648 hours, the Pacific Gas and Electric Company (PG&E) received a
Supervisory Control and Data Acquisition (“SCADA”) alert regarding the Buck Creek 1101
Circuit.1 The alert indicated that there was an outage at Cresta Dam, which is serviced by Bucks
Creek 1101 Circuit. PG&E dispatched a troubleman (Dixie Troubleman) to investigate, who
arrived at the dam at 1218 hours. The Dixie Troubleman observed blown fuses located on a pole
up the hill from Cresta Dam. The Dixie Troubleman did not identify the reason the fuses blew
while at Cresta Dam. When the Dixie Troubleman arrived at the location of the fuses at 1650
hours, he observed two of the three fuses blown, a tree resting on the circuit down the hill from
the fuses, and a small fire. The Dixie Troubleman attempted to fight the fire in addition to
radioing for help. The California Department of Forestry and Fire Protection (CAL FIRE) was
contacted.
CAL FIRE determined that the fire was caused by a tree contacting electrical distribution lines
owned and operated by PG&E.2 The tree fell and hit the lines, which caused two of the three
lines to become electrically connected, resulting in a phase-to-phase fault. 3 This fault blew two
1
SCADA is the system that PG&E uses to remotely monitor and control its electric circuits.
2
CAL FIRE, “CAL FIRE NEWS RELEASE – CAL FIRE Investigators Determine Cause of the Dixie Fire” (CAL
FIRE News Release (January 4, 2022), page 1.
3
CAL FIRE, “CAL FIRE Investigation Report – Case Number: 21CABTU009205-58 – Dixie” (CAL FIRE
Investigation Report) (July 13, 2021), page 45.
Page 1 of 50
fuses. Even though two fuses operated and deenergized the lines, the third conductor remained
energized and in contact with the tree, which caused a high-impedance fault. The energized line
in contact with the tree eventually started a fire.
The Safety and Enforcement Division’s (SED) investigation found that PG&E violated
requirements of the California Public Utilities Commission’s (CPUC) General Order (GO) 95
and the California Public Utilities Code (PU Code), as listed below:
Public Utilities Code Section 451 PG&E failed to adequately consider the
hazard of Bucks Creek 1101 circuit in its
response to the outage at Cresta Dam
Page 2 of 50
B. Witnesses
Name Title
1 Henry Sweat CPUC Lead Investigator
2 Sam Mandell CPUC Investigator
5 PG&E Incident Investigator
6 PG&E Claims Supervisor
8 PG&E Arborist
9 Matt Palades CAL FIRE Lead Investigator
10 Shawn Zimmermaker CAL FIRE Battalion Chief
11 Dixie Troubleman PG&E Troubleman
12 NDCC Operator #1 PG&E Distribution Operator
13 PG&E Hydroelectric Operator PG&E Hydroelectric Operator
14 PG&E Roving Operator PG&E Roving Operator
Page 3 of 50
C. Evidence
Source Description
1 PG&E Initial Online Incident Report, 7/18/21
2 CPUC Field Investigation, 7/26/2021
Case 3:14-cr-00175-
3 Documents 1408 - 1532, 7/28/2021 through 12/08/21
WHA
4 CPUC Data Request #1, 8/4/2021
5 PG&E Responses to Data Request #1, 8/6/2021 through 9/1/2021
6 CPUC Data Request #2, 8/10/2021
7 KCBS News report, 8/12/2021
8 PG&E 20-day report, 8/13/2021
9 PG&E Responses to Data Request #2, 8/16/2021 through 5/25/2022
10 CPUC Data Request #3, 8/16/2021
11 PG&E Data Request #4, 8/24/2021
12 CPUC Field Investigation, 8/25/2021
13 PG&E Responses to Data Request #3, 8/26/2021
14 PG&E Responses to Data Request #4, 9/24/2021 through 3/25/2022
15 CPUC Field Investigation, 10/14/2021
16 CPUC Field Investigation, 11/23/2021
17 CAL FIRE Press Release, 1/4/2022
18 CPUC Data Request #5, 1/31/2022
Page 4 of 50
II. Background
SED reviewed and analyzed records, examined physical evidence, and interviewed witnesses
related to this incident to determine compliance with Commission rules and regulations,
specifically GO 95 and GO 165.4 SED conducted field observations of evidence collection and
reviews of PG&E’s operations and maintenance procedures and relevant records. SED submitted
six data requests totaling 125 questions to PG&E. The questions included requests for
procedures, records, forms, and responses to specific questions related to the Dixie Fire. SED
also reviewed CAL FIRE’s investigation report, associated exhibits, arborist report and photos.
The Dixie Fire started on July 13, 2021, just off Storrie Road near Cresta Dam in Butte County at
approximately 39.874608, -121.378855 (Figure 1 and Figure 2). The fire was contained on
October 26, 2021. The Dixie Fire burned 963,309 acres, destroyed 1,311 structures, damaged 94
additional structures and caused four injuries. CAL FIRE determined that the cause of the fire
was a tree contacting electrical distribution lines on the Bucks Creek 1101 12 kV circuit owned
and operated by Pacific Gas and Electric (PG&E).5 The CAL FIRE report summarized causation
of the fire as follows:
The fire ignited as a result of a 65’ tall, damaged and decayed Douglas fir tree when it fell
and contacted conductors at approximately 6:48 AM. Two of the three fuses blew (opened)
upon initial contact with the conductors, but the third fuse remained closed and kept a line
energized. The tree being in contact with energized conductors and the ground created a high
impedance fault. The high impedance fault energized the tree, which caused heat and arcing
to ignite a dry and receptive fuel bed over the course of 10 hours.6
Cresta Dam is located directly off Highway 70. While the ignition location of the fire is very
close to Cresta Dam geographically, the incident location is up a very steep hill. Driving to the
incident location from the dam involves an 18-mile circuitous drive, most of which is on a windy
dirt road.
4
This investigation did not assess whether PG&E complied with its Wildfire Mitigation Plans (WMP).
5
CAL FIRE Investigation Report, page 45; PG&E Electric Incident Report Form, 20-Day report (August
13, 2021) (20-Day report), page 1.
6
CAL FIRE Investigation Report, page 5.
Page 5 of 50
Figure 3: Sketch of fire size and location when the Dixie Troubleman arrived at the incident
location at approximately 1650 hours. Sketch has been cropped for clarity.9
9
PG&E, “Exhibit X-1” (August 25, 2021) (Exhibit X-1), page 2. Exhibit 1 was submitted to SED in
response to Data Request 1, Question 4.
Page 8 of 50
Figure 5: Subject Tree leaning on the Bucks Creek 1101 circuit.12 The photo was taken on July
18, 2021, standing near Pole 100403908 looking downhill toward Cresta Dam and Pole
100403909.
12
PG&E. Photo: “2021-07-18_0045,” attached to 20-day report.
Page 10 of 50
Figure 6: Subject Tree stump13
The weather on the day of the fire did not meet PG&E’s criteria for implementation of a Public
Safety Power Shutoff since there was no wind event forecasted and no Red Flag Warning issued
by the National Weather Service.14 Table i, below, shows weather data recorded at the three
closest PG&E-owned weather stations to the incident location at 1700 hours on July 13, 2021:
Table i: Recorded weather data on July 13, 2021 at 1700 hours.15
Temperature Relative Wind Speed
Station Name Location
(F) Humidity (%) (miles/hour)
PG326 39.82864, -121.47166 84.7 19.91 11.2
PG468 39.76488, -121.48608 89.8 18.74 5.89
PG328 39.79947, -121.48370 92.4 16.55 3.63
13
PG&E. Photo: “2021-07-18_1285,” attached to 20-day report.
14
PG&E Response to Data Request Dixie Fire-SED-004, Question 40 (September 24, 2021), page 1.
15
PG&E “# Station: PG326” (PGE-DIXIE-CPUC-000001015); “# Station: PG468” (PGE-DIXIE-CPUC-
000001016); “# Station: PG328” (PGE-DIXIE-CPUC-000001017) (July 13, 2021).
Page 11 of 50
III. SED Review and Analysis
This section describes SED’s investigation, which includes a detailed event timeline, field
observations, and a review of relevant documents from PG&E.
A. Event Timeline
The following is a timeline of the events of July 13, 2021, unless otherwise noted.16
1. 0648 – Line Recloser 1101/2 recorded current levels on two of the three Bucks Creek
1101 phase exceeding the Minimum to Trip threshold. The duration was less than 4/100s
of a second and did not meet the minimum time length to trip the recloser.
2. 0721 (approximate, exact time unknown) – In response to a SCADA alert indicating loss
of station power at Cresta Dam, PG&E Hydroelectric Operator asked a PG&E Roving
Operator (Rover) of the outage alarms to investigate.
3. 0721 – Rock Creek Operator informed an operator at PG&E’s Northern Distribution
Control Center (NDCC) of the outage
4. 0852 – The Rover reported to Rock Creek Operator that station service was out at Cresta
Dam and the lights were out in the tunnel next to the dam.
5. 0904 – The Rover concluded to the Rock Creek Operator that the outage was between the
Bucks Creek Substation and Cresta Dam. This was reported to the NDCC at 0911 hours.
The distribution operator at the NDCC stated they would dispatch a troubleman to
investigate.
6. 0936 – A PG&E Dispatcher created a non-emergency Priority One field order (or “tag”)
for a troubleman (Quincy Troubleman) to investigate the outage. The Quincy
Troubleman assigned to investigate the tag from Quincy, CA responded that the tag was
not in his response area and should have been assigned to a troubleman located in Chico
or Paradise.
7. 1047 – PG&E Dispatcher assigned tag to the Dixie Troubleman.
8. 1053 – Dixie Troubleman reported he was in route but stopped on the way to address
another Priority One tag.
9. 1125 – Dixie Troubleman called the distribution operator at NDCC, who reported they
were still seeing power reading on Bucks Creek 1101 from Line Recloser 1101/2. The
NDCC Operator directed the Dixie Troubleman to check fuses on the circuit sources side
of Cresta Dam: Fuse 805 and Fuse 17733. Fuse 805 is near Cresta Dam. Fuse 17733 is on
the source side of Fuse 805 and is attached to Pole 100403908.
10. 1218 – Dixie Troubleman arrived at Cresta Dam. He observed Fuse 805 had not operated.
He observed that the meter being served by the transformer one span upstream of Fuse
805 was de-energized. He patrolled the area to investigate the cause of the power outage
but did not observe anything at Cresta Dam. Using binoculars, he visually inspected the
portion of Bucks Creek 1101 running between Cresta Dam towards Fuse 17733. All the
poles and wires that he could see on that circuit appeared to him to be up and in their
16
The source of the timeline information is PG&E’s 20-Day report for the Dixie Fire, unless noted otherwise.
Page 12 of 50
normal position, not bent or twisted. He saw what appeared to be an open fuse cutout at
Fuse 17733 hanging down from Pole 100403908. This would indicate that one of the
fuses at Fuse 17733 had operated. He did not see any vegetation on the circuit or any
smoke or other indication of a fire. The Troubleman decided to drive to investigate Fuse
17733.
11. 1327 – Dixie Troubleman arrived at bridge two miles from Pole 100403908 and Fuse
17733. The bridge was under repairs. The Troubleman was informed it would be about
two hours before he could pass. After speaking with the construction crew, he turned
around.
12. 1500 (approximate) –Dixie Troubleman returned to Highway 70 to a location with cell
service. He saw that he had received two priority zero emergency tags. He contacted the
NDCC operator to ask if he should attend the priority zero tags. The NDCC operator
advised that other troubleman were closer to the priority zero tags and that the Dixie
Troubleman should address the outage at Cresta Dam. The Troubleman drove a different
route from Highway 70 that reconnects with Storrie Road before the closed bridge on this
trip to Pole 100403908 and Fuse 17733.
13. 1500 (approximate) – A Senior Power Generation Inspector thought he smelled and saw
smoke southeast of Bucks Creek Powerhouse.17 The inspector called the Rock Creek
Switching Center and alerted the operator there that he smelled smoke. The operator
alerted a PG&E helicopter pilot and asked the pilot to fly over the Bucks Creek area to
look for smoke or signs of a fire. The PG&E helicopter pilot immediately rerouted from
its current flight plan and flew over the Bucks Creek area. The pilot did not see a fire or
any indication of fire during his flight over the Bucks Creek area.18 The operator also
asked an electrician working outside the Rock Creek Powerhouse if he smelled smoke.
The electrician circled the facility; he did not smell smoke or see any sign of fire.19
14. 1630 – Dixie Troubleman returned to bridge.
15. 1650 – Dixie Troubleman arrived at Pole 100403908 and Fuse 17733. Dixie Troubleman
observed two of three fuse cutouts open. He opened the third fuse. Before he opened the
fuse, he observed a fire 60-80 yards downhill from his position, approximately 600 to 800
square feet in size. He also observed a tree leaning on the span between Pole 100403908
and Pole 100403909.
16. 1701 –A PG&E employee at the Rock Creek Powerhouse heard a radio call by a PG&E
employee driving south on Highway 70 that observed a small plume of smoke.20 CAL
FIRE was notified at 1706 hours. The Dixie Troubleman attempted to put fire out. He
then returned to his truck and spoke to his supervisor on the radio. After the conversation,
he continued to fight the fire.
17
Cresta Dam and the ignition area are southwest of the Bucks Creek Powerhouse.
18
PG&E. “Response to Data Request Dixie Fire-SED-005, Question 18,” (February 28, 2022), page 1.
19
“Document 1479 - Responses to Orders for Further Responses Re Dixie Fire” (September 24, 2021)
(Document 1479), page 10.
20
PG&E. “Dixie Factual Summary,” (September 2, 2022), page 4.
Page 13 of 50
17. 1730 – The Dixie Troubleman observed a CAL FIRE spotter plane followed by a CAL
FIRE helicopter and fixed-wing aircraft for fire suppression.
18. 1900 – A CAL FIRE ground crew arrived.
19. 2000 – The Dixie Troubleman left the scene.
20. 2030 – PG&E de-energized Bucks Creek 1101 at Switch 941 at the request of CAL
FIRE.
21. July 14, 2021 at 2000 – PG&E de-energized the entire Bucks Creek 1101 circuit at the
request of CAL FIRE.
21
Exponent is a consultant hired by PG&E.
22
For a map showing the incident location and pole locations relative to Storrie Road, see Figure 2 of this
report.
Page 14 of 50
Figure 9: One of the insulators from Pole Figure 10: South phase conductor from the
100403909 incident span
SED viewed Pole 100403909 and the surrounding area. Pole 100403909 appeared to be in sound
condition and no obvious signs of fire were present. Copper insulators were observed near the
pole where CAL FIRE had previously removed the conductor. CAL FIRE had previously
collected a portion of the Subject Tree. The remaining portion of the tree was located near the
pole (Figure 11 and Figure 12). The tree did not appear to be burned, and signs of ignition were
not apparent on the tree.23 There was a cluster of roots from the stump removed by CAL FIRE
estimated to be 40 feet from Pole 100403909 (Figure 13). This root cluster was the base of the
Subject Tree. The root shown in Figure 14 had a significant amount of rot at the center.
23
CAL FIRE had already removed the sections of the Subject Tree showing burns or signs of ignition
when SED was on site.
Page 16 of 50
Figure 11: Base of Pole 100403909 showing fallen Figure 12: Subject Tree
Subject Tree
Figure 13: Cluster of roots Figure 14: Root showing significant rot
Page 17 of 50
The purpose of the site visit was to view the evidence CAL FIRE collected from the ignition site
of the Dixie Fire. The following is a description of the evidence collected and related
observations:
Conductors
Multiple conductors showing signs of tree contact
Conductors typically had fibers from the tree embedded between the individual wires
of the conductor (Figure 15).
Two conductors showing signs of arcing and phase-to-phase contact (Figure 16)
Fuses (3)
Two fuses operated, one did not operate (Figure 17 and Figure 18)
The two fuses that operated appeared to have operated correctly.
Fuse Cut Outs: No signs of misfiring fuses.
Jumpers: No signs of arcing.
Subject Tree
Burned portion of the trunk that was in contact with the energized conductor (Figure
19)
A tree limb that had folded up and contacted the energized conductor
Sections of the Subject Tree showing signs of decay. A segment of the tree where it
broke off from the stump was severely burned (Figure 20).
Subject Tree stump with burn marks and decay (Figure 21).
Figure 15: Energized conductor in contact Figure 16: Signs of arcing from phase-to-
with tree phase contact on conductor
Page 18 of 50
Figure 17: One of two fuses that operated Figure 18: Fuse that did not operate
Figure 19: Subject Tree trunk where energized Figure 20: Section of the Subject Tree trunk
conductor contacted tree. that failed causing the tree to fall.
Page 19 of 50
Figure 21: Subject Tree stump. A large portion of tree trunk was damaged before the fire started.
Cresta Dam
From the CAL FIRE evidence locker in Oroville, SED representatives continued to Cresta Dam.
The goal of the visit was to view, from the dam, the segment of the Bucks Creek Line where the
fire originated. 24 Both Pole 100403909 that was closest to the fallen tree and Pole 100403908
that supported the fuses were visible from Cresta Dam without binoculars. With binoculars,
features of the two poles and the line were easily distinguishable. SED observed that due to the
positioning of the fuses, it would likely have been difficult to accurately determine how many
fuses had blown. As of the date of the site visit, PG&E had not replaced the conductor between
the two poles but had installed a single cable between the two poles. The cable was visible over
the entire span between the two poles (Figure 22).
24
The Dixie Troubleman visually inspected the line from Cresta Dam on July 13, 2021 and had reported
that “the fuse may have tripped on at least one of the three phases of the line. At that point, there was no
vegetation seen on the line, nor any smoke or other indication of fire.” See Document 1408-1,
Declaration of in Support of Response to Order Requesting Information on Dixie and Bader
Fires (July 28, 2021) (Document 1401-1), page 3. (Name of Declarant redacted in original.)
Page 20 of 50
C. Document Review and Investigation
SED submitted six data requests totaling 125 questions to PG&E. The questions included
requests for procedures, records, forms, and responses to specific questions related to the Dixie
Fire. The questions loosely consisted of the following categories: Inspections/Work Orders,
Vegetation Management, System Operation and Additional Documents. Additionally, SED
reviewed documents generated by the Federal Case 3:14-c-00175-WHA in the Northern
California District Court (District Court Case).27
1. Inspections and Work Orders
SED reviewed the five most recent patrol inspections and the three most recent detailed
inspections on the portion of the circuit spanning five structures in both direction from the
incident area. SED finds PG&E in violation of GO 95, Rule 18 for failure to complete Electric
Overhead Tag 109671451 within the twelve-month corrective action deadline of October 30,
2015.
Inspections
The last two detailed inspections performed per GO 165 were in December 2016 and May 2021.
PG&E did not identify required work at Pole 100403908 or Pole 100403909, or the adjacent
spans.28 In June 2019 and May 2020, PG&E performed patrols per criteria specified in GO 165.
PG&E did not identify any items for related to Pole 100403908 or Pole 100403909.29 PG&E
also performed a Wildfire Safety Inspection Program inspection in May 2019, which is similar to
a detailed inspection. PG&E identified Pole 100403908 as damaged and replaced the pole on
July 21, 2019.30
Work Orders
There were two late work orders for poles within five spans of the immediate area of interest.
The first was the work order for Electric Overhead Tag 109671451.31 PG&E identified the scope
of work originally as a crossarm replacement but changed the scope to a replacement of the
entire pole due to decay.32 A PG&E representative identified the corrective action on October
27
Generally, the District Court Case filings were generated in response to questions posed by Senior
District Judge William Alsup.
28
PG&E “Electric Maintenance Inspection Log” (December 12, 2016), page 1; “100403908 2021 OH
Checklist (May 13, 2021), pages 1-5; “100403909 2021 OH Checklist” (May 13, 2021), pages 1-5.
29
PG&E “Electric Maintenance Patrol Log” (December 6, 2019) page 1; “Electric Maintenance Patrol
Log” (June 10, 2020), page 1.
30
PG&E “NV-Electric Distribution Overhead Inspection – 18105179916” (May 19, 2019), page 6;
“Electric Overhead Tag Notification #117385786” (EC Tag #117385786) (Identified June 6, 2019, latest
comments added August 16, 2019), page 1.
31
PG&E “Electric Overhead Tag Notification #109671451” (EC Tag #109671451), Date Identified
October 30, 2014. Latest comments added February 21, 2017.
32
EC Tag #109671451.
Page 24 of 50
30, 2014 and set a due date of October 30, 2015.33 The work was completed on November 16,
2016.34 At the time the Electric Overhead Tag 109671451 was created, GO 95, Rule 18 required
the time limit for corrective action to be determined by a qualified company representative, with
a maximum time limit of 12 months when related to work safety and 59 additional months for all
other purposes.35 Because PG&E’s qualified company representative determined a deadline of
one year, the GO 95, Rule 18 compliance period is one year. In Data Request 5, Question 30,
SED asked PG&E to provide a justification for the late work that would meet the requirements
permitted by GO 95 Rule 18.A.2.b, which allows for extension of correction times under
reasonable circumstances, such as permitting issues.36 PG&E did not provide a justification.37
The second work order was for Electric Overhead Tag 117162503.38 PG&E determined a pole
replacement was required. PG&E identified the issue on May 4, 2019 and set the due date for
June 30, 2019. The plans were submitted to Plumas County on May 28, 2019 but did not receive
clearance until July 23, 2019. PG&E completed construction on October 3, 2019. With regards to
the late completion, PG&E stated the following: “Electric Overhead Tag 117162503 was
completed late due to delay in third-party environmental review” and “We believe that this
justification satisfies the standard set forth by of GO 95, Rule 18.A.1.b, which states that
correction times may be extended under reasonable circumstances such as permit requirements
and third party [sic] refusal.”39
2. Inspections and Work Orders Analysis
PG&E failed to complete the Electric Overhead Tag 109671451 within the one-year deadline
determined by a qualified company representative. Therefore, SED finds PG&E in violation of
GO 95, Rule 18.
SED accepts the justification for Electric Overhead Tag 117162403, and agrees that GO 95, Rule
18.A.2 allows for an extension under reasonable circumstances such as permit requirements. The
tag was completed after five months, which is beyond PG&E’s initial internal deadline of eight
weeks. While PG&E exceeded their internal deadline, GO 95, Rule 18.B.1.a.ii permits 12
months to complete Level 2 corrective actions. Taking the environmental permitting review into
consideration, SED does not find a violation of GO 95, especially considering the corrective
action was completed before the 12-month deadline.
33
EC Tag #109671451
34
EC Tag #109671451
35
California Public Utilities Commission, “Decisions Adopting Regulations to Reduce Fire Hazards
Associated with Overhead Power Lines and Communication Facilities,” Page B-4. January 12, 2012.
36
PG&E Response to Data Request Dixie Fire-SED-005, Question 30 (March 25, 2022), page 1.
37
Response to Data Request Dixie Fire-SED-005, Question 30, page 1.
38
EC Tag #117385786
39
Response to Data Request Dixie Fire-SED-005, Question 30, page 1.
Page 25 of 50
3. Vegetation Management
This section discusses PG&E vegetation management practices and the vegetation patrols that
could have identified and removed the Subject Tree. SED finds PG&E in violation of GO 95,
Rule 31.1 for failing to update its procedures to meet the requirements of GO 95, Rule 35. SED
also finds PG&E in violation of GO 95, Rule 18.B for failing to maintain records showing (1) the
correct date of inspection for 2020, and (2) the areas that were inspected in 2019. Finally, SED
finds PG&E in violation of GO 95 for failing to identify the Subject Tree as hazardous and thus
failing maintain the 12kV overhead conductors safely and properly with accepted good practice.
Procedures
PG&E’s vegetation management strategy is comprised of multiple programs. Two of the most
relevant are the routine vegetation management program and the second patrols that include
Catastrophic Event Memorandum Account (CEMA) patrols.40
The primary goal of the vegetation management program is to ensure safe and reliable operation
of facilities and prevent foreseeable vegetation-related outages. In order to implement these
goals, PG&E created a multitude of procedures including “Distribution Vegetation Management
Standard” (DVMS),41 “Distribution Routine Patrol Procedure” (DRPP),42 “Vegetation
Management Second Patrol Procedure,”43, “Reporting Abnormal Field Conditions Procedure,”44
”Vegetation Management Priority Tag Procedure,”45 and “Facility Protect and Work Difficulty
Classification Procedure.”46 The procedures implement the goals of the vegetation management
program by requiring an annual patrol that identifies tree work on all overhead facilities. The
annual patrol identifies hazard trees and trees that are either encroaching or are likely to encroach
on the minimum distance requirements (MDRs) or minimum clearance requirements.
The MDRs from vegetation to distribution assets, such as conductors, are set by GO 95, Rule 35
at 1.5 feet generally and four feet for vegetation in High Fire Threat Districts (HFTDs). The
PG&E Standards, DVMS and DRPP, set their MDRs as the MDRs specified in Appendix A,
which is attached to both documents. Appendix A states that the MDR as required by GO 95,
40
A second patrol is used to reduce the vegetation risk in high hazard areas by performing a patrol
approximately six months after a routine patrol. The goal of the patrol is to identify hazardous vegetation.
41
PG&E “Distribution Vegetation Management Standard (DVMS)” (September 4, 2015).
42
PG&E “Distribution Routine Patrol Procedure (DRPP)” (October 27, 2015).
43
PG&E “Vegetation Management Second Patrol Procedure” (PG&E Procedure: TD-7102P-23) (July 31,
2019).
44
PG&E “Reporting Abnormal Field Conditions Procedure” (PG&E Procedure: TD-7102P-09)
(November 24, 2014).
45
PG&E “Vegetation Management Priority Tag Procedure” (December 26, 2021).
46
PG&E “Facility Protect and Work Difficulty Classification Procedure” (April 1, 2015).
Page 26 of 50
Rule 35 is 1.5 feet generally and four feet in Santa Barbara County.47 The MDRs in Appendix A
reference a superseded version of GO 95.48
PG&E frequently does ad-hoc updates of its standards through bulletins. PG&E issued “TD-
7102B-015, High Fire-Threat District Bulletin” on February 2, 2018. While this bulletin was
active, it superseded the Appendix A MDRs referenced by the DRPP and DVMS.49 The bulletin
states that 48 inches of clearance is always required between vegetation and distribution voltage
within the HFTD.50 PG&E retired this bulletin on August 26, 2020, so the requirements of the
bulletin no longer apply. PG&E provided no updates to Appendix A in effect for the period after
August 26, 2020.51 PG&E stated that the Vegetation Management Second Patrol Procedure
requires an MDR of four feet, but this procedure only applies to second patrols.52
Another goal of PG&E’s vegetation management program is to identify and ameliorate Hazard
Trees.53 After identifying a Hazard Tree, the inspectors perform at least a Level One (limited
visual) inspection as defined by the International Society of Arboriculture.54
GO 95 Rule 18.B requires that PG&E maintain an auditable maintenance program for its
facilities and lines. To track the progress of vegetation management inspections, PG&E uses
index maps signed by the vegetation management inspector to indicate that the map area was
inspected. The map labeled AT112-C was missing from the 2019 vegetation management
inspection records.55 As of July 27, 2022, PG&E was not able to locate this record.56
Review of Recent Vegetation Inspections and Information Regarding the Subject Tree
SED reviewed the documentation for the last five vegetation management inspections starting in
2016. The following is a timeline of the inspection cycle for 2020 and 2021:
March 5, 2020 – CEMA vegetation management patrol.57
47
DRPP, page 19.
48
GO 95 was updated by CPUC Decision 17-12-024 in 2017 to mandate a MDR of four feet in all
HFTDs, but PG&E’s Appendix A was never updated accordingly.
49
PG&E Response to Data Request Dixie Fire-SED-006, Question 3 (July 27, 2022), page 1.
50
PG&E, “High Fire-Threat District Bulletin,” page 1 (February 15, 2018).
51
Response to Data Request Dixie Fire-SED-006, Question 3, page 1.
52
Response to Data Request Dixie Fire-SED-006, Question 3, Page 1.
53
“Hazard Trees” are defined by PG&E’s DVMS as “Trees that are dead, show signs of disease, decay or
ground or root disturbance, which may fall into or otherwise impact the conductors, towers or guy wires
before the next inspection cycle”
54
PG&E Response to Data Request Dixie Fire-SED-005, Question 23, page 1 (March 14, 2022).
55
PG&E Vegetation Management Bucks Creek 1101 2019 Index Map (November 12, 2019).
56
PG&E Response to Data Request Dixie Fire-SED-006, Question 4 (July 27, 2022), page 1.
57
PG&E Response to Data Request Dixie Fire-SED-005, Question 23, page 1.
Page 27 of 50
December 3, 2020 – Date of annual routine patrol of the circuit. PG&E internal
documentation (called an Index Map) reflects a date of completion on November 11,
2020, but SED understands the correct date is December 3, 2020.58 PG&E did not know
why the correct inspection date was not reflected on the index map.59
January 14, 2021 – CEMA vegetation management patrol performed via helicopter. 60 No
work beyond the 2020 routine patrol was identified.61
June 5, 2021 – Vegetation work finished as required by the routine patrol on December 3,
2020.62 PG&E is not aware of a regulatory required due date to perform the work after it
is identified.63
July 1, 2021 – Originally planned completion date of routine vegetation management
patrol for 2021. The contractor was responsible for completing it within 60 days of the
planned date. The patrol was scheduled for completion on July 23, 2021, but this did not
occur due to the fire starting on July 13, 2021.64
As part of its routine vegetation management patrols and work, PG&E generates a document
called “Inspection Record Detail.”65 The Inspection Record Detail is a running list of all the trees
worked along a specific length of the circuit since the record was created.66 The Inspection
Record Detail lists 11 trees total, three of which are trees on the span of line between Pole
100403909 and Pole 100403908: a true fir (labeled Tree 7) originally classified for removal but
not removed; a Ponderosa pine (labeled Tree 8) that was trimmed on September 6, 2012; and a
Ponderosa pine (labeled Tree 9) removed on September 6, 2012.67
As previously noted, the Subject Tree was a Douglas fir. PG&E’s Inspection Record Details
going back to 2016 do not mention the Subject Tree. PG&E provided photos of the Subject Tree
taken in 2019 (Figure 27), in which the Subject Tree’s canopy appeared green.68 As described in
SED’s Field Observation in section III.B.1, at least one of the eight roots had internal rot. SED
58
PG&E Vegetation Management Bucks Creek 1101 2020 Index Map (November 11, 2020).
59
PG&E Response to Data Request Dixie Fire-SED-006, Question 1, (July 27, 2022), page 1.
60
PG&E Response to Data Request Dixie Fire-SED-005, Question 23 (February 28, 2022), page 1.
61
CN Utility Consulting, Inc. Document 1515-9: Declaration of in Response to Nov. 3, 2021
Order (November 16, 2021) (Document 1515-9), page 3. (Name of Declarant in redacted original.)
62
PG&E Response to Data Request Dixie Fire-SED-005, Question 25 (March 14, 2022), page 1.
63
PG&E Response to Data Request Dixie Fire-SED-005, Question 25 (March 14, 2022), page 2.
64
PG&E Response to Data Request Dixie Fire-SED-005, Question 23 (February 26, 2022), page 1.
65
PG&E. “Response to Data Request Dixie Fire-SED-005, Question 24,” Page 1. March 14, 2022.
66
PG&E. “Response to Data Request Dixie Fire-SED-005, Question 24,” Page 1. March 14, 2022.
67
PG&E. “Inspection Record Detail,” Page 1 and 2. November 17, 2020.
68
PG&E. “Response to Data Request Dixie Fire-SED-004, Question 60,” Page 1. September 24, 2021.
The color of the canopy is a key factor in vegetation management inspectors’ determinations of tree
health.
Page 28 of 50
Figure 27: The yellow arrow points to the tree that fell on the circuit and started the
Dixie Fire.73
Tree workers who performed the patrols testified in the District Court Case about their process.74
These workers included the consulting utility forester (CUF), who performed the inspection on
December 3, 2020, and the senior consulting utility forester (SCUF) who both audited the CUF’s
patrol and performed the CEMA inspection on January 14, 2021.75 The CUF testified that they
follow the International Society of Arboriculture (“ISA”) Utility Tree Risk Assessment Best
Management Practices, which identifies three levels of assessment.76 Most inspections are Level
One or Limited Visual assessments, which are used to identify specific conditions or obvious
defects.77 A Level Two or Basic assessment requires the inspector to walk around the tree and
73
Pacific Gas and Electric. “Document 1416 - Exhibit P-5.” Photo taken in 2019.
74
See Document 1517-7: Declaration of in Response to the Court’s November 3, 2021 Order and
Document 1515-9: Declaration of in Response to Nov. 3, 2021 Order. (Name of Declarant
redacted in original.)
75
Document 1517-7, Declaration of in Response to the Court’s November 3, 2021 Order
(November 16, 2021) (Document 1515-7), page 2; Document 1515-9: Declaration of in
Response to Nov. 3, 2021 Order (November 16, 2021) (Document 1515-9), page 2. (Name of Declarant
redacted in original.)
76
Document 1515-7, page 2.
77
Document 1515-7, page 2.
Page 30 of 50
look at the roots, trunk, and branches.78 The CUF stated that a Level Two assessment is used
when causes of concern are noticed during a Level One assessment.79 After reviewing photos of
the tree, including Figure 27 on page 30, the CUF made the following statements, among others:
The CUF did not recall the Subject Tree in the photos, but it appeared to be a healthy fir80
The CUF would have performed a Level One assessment of the tree by walking
underneath the powerlines in the right-of-way and observing those trees for signs of
disease distress or structural or compliance problems81
The CUF did not recall performing a Level Two assessment of the Subject Tree. 82
The CUF did not see any indications from the photos that Level Two assessment should
have been performed83
The CUF also stated that inspecting a tree’s root structure is generally outside of the scope of
inspections performed since the roots are buried.84 The SCUF who audited the CUF’s routine
patrol for Bucks Creek 1101 circuit testified that they did not audit the portion of the patrol near
the Incident Location.85 The same SCUF conducted the CEMA patrol of Bucks Creek 1101 from
a helicopter on January 14, 2021.86 The SCUF observed two to three trees that required
trimming, none of which were located near the Incident Location and all of which were noted
during the routine patrol on December 3, 2020.87 The SCUF also stated that based on review of
the photos, it was unlikely that the SCUF would have identified the Subject Tree as requiring
trimming or removal.88
78
Document 1515-7, page 2. A Level Three is performed rarely and is an advanced assessment.
79
Document 1515-7, page 3.
80
Document 1515-7, page 3.
81
Document 1515-7, pages 3-4.
82
Document 1515-7, page 5.
83
Document 1515-7, page 5.
84
Document 1515-7, page 4.
85
Document 1515-9, page 2.
86
Document 1515-9, page 2.
87
Document 1515-9, page 3.
88
Document 1515-9, page 3.
Page 31 of 50
CAL FIRE Arborist Report
The CAL FIRE Arborist Report (Arborist Report) describes extensive decay and defects at the
base of the Subject Tree, as a result from 2008 Butte County Fire, and an unidentified event
between 2015 and 2016.89 The report states:
In my opinion the degraded condition of the base of the Douglas fir was the primary
cause of failure of the tree and that defect would have been visible without extraordinary
discovery effort, from under the conductors. It would have been visible as a catface, an
open injury on the trunk. A cursory visual inspection around the base of the tree would
have revealed the poor mechanical condition of the tree. A pre-inspector who was close
to the tree for a brief visual inspection for any reason should have discovered the decay.90
Above the roots evidence suggest that the south side of the lower trunk was burned and
killed over more than half its circumference. This would have formed a wound with no
bark cover, visible as exposed and decaying wood. The extent of this would is well-
illustrated in Figures 35 and 36. It is my opinion that most of the sound wood supporting
the tree is represented graphically in Figure 36. The absence of sound wood where
expected elsewhere is dramatic.91
Figure 28 below shows “Figure 36” from the Arborist Report. In discussing Figure 36, the
arborist states the following:
Nearly all the wood in that column to the south toward the conductors is missing. It is my
opinion that the missing wood through this stump was in a state of advanced decay, as
seen higher in the tree trunk in Figure 8, and was either consumed complete by decay
organisms or was so susceptible to combustion that it quickly burned in the 2021 fire.92
89
McNeil Arboriculture Consultants LLC. “Exhibit W: Arborist Report,” (October 10, 2021) (Arborist
Report), page 20.
90
Arborist Report, page 21.
91
Arborist Report, page 21.
92
Arborist Report, pages 20 and 21.
Page 32 of 50
Summary of Vegetation Management-Related Violations
96
PG&E Vegetation Management Bucks Creek 1101 2020 Index Map (November 11, 2020).
97
PG&E Vegetation Management Bucks Creek 1101 2019 Index Map (November 12, 2019).
98
A detailed timeline of events on July 13, 2021 is provided in Section III.A of this report.
99
PG&E Response to Data Request Dixie Fire-SED-005, Question 2 (March 14, 2022) page 1.
Page 34 of 50
continuous SCADA monitoring of the Bucks Creek 1101 circuit, checking load and circuit
information.100 Distribution Operators are authorized to deenergize distribution lines to
protect public safety and property.
Hydro Operator #1 (the Hydro Operator who first called NDCC Operator #1 to discuss the
outage at Cresta Dam) – The Hydro Operator is like a Distribution Operator, but for
hydroelectric operations such as dams.
Hydroelectric Roving Operator (Rover) – A Rover’s primary responsibilities include
checking station service, confirming that stand-by generators were running if there was a loss
of power, and reporting his observations to the Hydro Operator at the Rock Creek Switching
Center.101 The Roving Operator was not responsible for identifying the source of the outage
on the Bucks Creek 1101 Line.102
b. PG&E Standard Procedures for Responding to Outages
PG&E Distribution Operators are responsible for the operation of PG&E’s electric distribution
system, including continuous monitoring of the Bucks Creek 1101 circuit.103 After learning about
an outage, a Distribution Operator reviews SCADA data and any other available information.
SCADA is the system used to communicate with devices on a circuit, recording and sending data
back and forth between the devices and the Distribution Operators. More specifically, when
outages occur Distribution Operators note whether the circuit breaker is open, whether the load
appears normal and balanced across all three phases, whether there is excessive ground current,
and any other pertinent information, e.g., known hazards on the line like a tree or an active fire.
The Distribution Operator may decide to contact PG&E Dispatch to send out a troubleman to
further assess and address trouble on the circuit. The troubleman’s job is generally to determine
the cause of the outage, ensure that the area and PG&E facilities are safe, and restore power to
customers. The troubleman is required to communicate with the Distribution Operator before
deenergizing the lines and is otherwise not permitted to de-energize the lines unless a safety
hazard requires prompt action.104
When the Distribution Operator receives a SCADA alert, the alert has a priority level. The top
level is Priority 10, reserved for fire detection.105 Priority 10 requires immediate response and
action. Priority levels P06-P09 are critical levels requiring immediate action, reserved for
“circumstances that indicate potential loss of equipment, path interruption, or customer
100
PG&E. Document 1515, Responses to Fifth Further Request for Responses Re Dixie Fire (September
24, 2021) (Document 1515), page 14. As noted previously, SCADA stands for “supervisory control and
data acquisition.”
101
PG&E Response to Data Request Dixie Fire-SED-006, Question 6 (July 27, 2022), page 1.
102
Response to Data Request Dixie Fire-SED-006, Question 6, page 1
103
Document 1515, page 14.
104
Response to Data Request Dixie Fire-SED-005, Question 2, page 2.
105
PG&E. “Responding to Emergencies and Alarms” (Utility Procedure: TD-2700P-09) (February 2,
2020), page 3.
Page 35 of 50
outage.”106 Priority levels P04-P05 are critical levels “comprised of all security alarms, other
significant substation trouble indicators, communication and field devices alarms.”107 At P04-
P05, however, immediate response is not required. The standard specifies further analysis and
action as needed. The decision to dispatch personnel or monitor the situation at P04-P05 is based
on “known concurrent activity and circumstances. In some instances, close monitoring may be
enough; in others, dispatching personnel may be required.”108
PG&E only de-energizes lines for known hazards.109 A PG&E Distribution Operator attested in
the District Court Case that fuses often operate or “blow,” so a “blown fuse” alone is not a
reason to deenergize the line.110 Fuses operate in response to faults or spikes in current and may
operate for a variety of reasons, often due to temporary issues that do not pose long-term safety
hazards. A fuse is designed to protect the rest of the system from fault events downstream of the
fault by cutting power downstream of the fuse, resulting in an outage.
Typically, distribution circuits have three phases. Each phase carries electricity, and normally
each phase has a fuse. In some cases, a fault can affect only one or two of the phases, causing
one or two of the fuses to operate. If two fuses operate to de-energize two of the three phases, the
third phase could remain energized. This is called single phasing, which can damage customer
equipment designed to use all three phases. Generally, a troubleman would open any remaining
fuses to prevent single phasing.111
When a troubleman arrives at a site, they inspect the area to determine the cause of the outage
and whether the line is safe to reenergize.112 Often, the troubleman walks the entire portion of the
deenergized circuit to make sure that there are no remaining hazards, such as a tree on the line.
PG&E considers a tree on a line a safety hazard requiring prompt action. Upon discovery of a
tree on the line, normally a troubleman would immediately deenergize the line.113
106
Utility Procedure: TD-2700P-09, pages 4-5.
107
Utility Procedure: TD-2700P-09, pages 4-5.
108
Utility Procedure: TD-2700P-09, pages 4-5.
109
PG&E Response to Data Request Dixie Fire-SED-001, Question 8 (August 31, 2021), page 1.
110
PG&E. “Document 1532-1 – Declaration of in Support of Response To Sixth Request re
Dixie Fire” (Document 1532-1), Page 2. September 17, 2021. (Name of Declarant redacted in original.)
111
Response to Data Request Dixie Fire-SED-005, Question 2, page 4.
112
Response to Data Request Dixie Fire-SED-005, Question 2, page 1.
113
Response to Data Request Dixie Fire-SED-005, Question 2, page 3.
Page 36 of 50
The SCADA alert received at 0648 hours on July 13, 2021 was assigned P04. The SCADA alert
indicated that the current measured by the recloser was above the minimum-to-trip (mtt).114 The
mtt is the lowest current value that PG&E has programed the recloser to operate and turn off the
power. The first alarm was followed closely by another notification which indicate that the
current was below the mtt.115 These alarms meant the current was briefly measured above the
mtt levels before returning to normal.
The Distribution Operator requested that PG&E Dispatch send a troubleman to investigate the
outage.116 When dispatching a troubleman, PG&E Dispatch assigns a priority to the tag.117 The
priority for tags is different than the priority system for SCADA alarms. Priority 0 tags indicate
“Timely Emergency Response” and apply to safety issues requiring an immediate response, such
as fires, arcing/bare wires, and downed wires.118 In contrast, Priority 1 tags, or “Same Day
Response,” apply to non-emergency services to address issues such as outages of unknown
causes.119 PG&E Dispatch assigned a Priority 1 tag to the outage at Cresta Dam, indicating that
the outage necessitated a same-day response but was not considered an emergency when a tag
was assigned.120
When diagnosing trouble based on the SCADA alert, NDCC Operator #1 had multiple pieces of
information available to assist in safely operating the system.121 NDCC Operator #1 had access
to SCADA alerts, but the alert just indicates priority and does not state what type of fault
triggered the alarm.122 NDCC Operator #1 also had access to historical information through the
SCADA system.123 This system transmits data at intervals of 15 and 30 seconds that includes the
magnitude of electrical current (or load) on each phase and the calculated ground current.124 This
system only updates if the phase load changes by one amp from the last reported phase.125 If the
114
The log description for the alert stated: “(P04) 1-Paradise Bucks Creek CB 1101 lr above mtt is
ALARM. See “Chico DO Switching Center SCADA” (July 13, 2021).
115
Chico DO Switching Center SCADA log description for the alert stated: “(P04) 1-Paradise Bucks
Creek CB 1101 lr above mtt is NORMAL.”
116
PG&E, “Document 1474-5 – 07-13-21 – 0914 - NDCC Operator #1 - Dispatcher” (July 13, 2021)
(Document 1474-5), page 2.
117
PG&E Response to Data Request Dixie Fire-SED-004, Question 53 (September 24, 2021), page 1.
118
Response to Data Request Dixie Fire-SED-006, Question 53, page 1.
119
A less urgent priority for tags, Priority 2, requires a scheduled response at a future date. Priority 2 is
used for nonemergency services such as discontinuing gas and/or electric service.
120
20-Day report, page 2.
121
PG&E Response to Data Request Dixie Fire-SED-005, Question 8, (March 14, 2022) pages 1-3.
March 14, 2022.
122
Response to Data Request Dixie Fire-SED-005, Question 8, page 1.
123
Response to Data Request Dixie Fire-SED-005, Question 8, page 1.
124
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
125
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
Page 37 of 50
data changes, the data is saved in the PI Historian database.126, 127 The PI Historian database
allowed NDCC Operator #1 to compare present load data to historical load data.128 NDCC
Operator #1 references reviewing the SCADA data on July 13, 2021 and stated the load data
appeared normal.129 SCADA data from July 13, 2021 is graphed and discussed in the attached
Technical Addendum; no violations are identified associated with this data.
The Dixie Troubleman testified that he could not turn off Switch 941 because he was worried it
130
would impact on downstream customers on the circuit like the On July 13, 2021,
NDCC Operator #1 had access to PG&E’s Digital Mapping System (DMS) to assist with
understanding the circuit on July 13, 2021.131 DMS displays maps and information regarding the
circuits. This information includes customer information such as the name on the account,
account number, SmartMeter number, service address, and sometimes a contact phone
number.132 DMS does not include information regarding customer equipment, such as back-up
batteries, unless it is a critical customer such as a hospital.133 For the Bucks Creek 1101 circuit,
there are three customers beyond Switch 941: Cresta Dam, the and
134
. PG&E stated that are not considered critical customers,
and that it had not identified any records indicating awareness as of July 13, 2021 that the
had battery backup power.135 The did, in fact, have battery backup.136 As a
result, out of the three customers connected to the portion of the circuit downstream of Switch
941, two lost power due to the operation of Fuse 17733, while the third, which did not lose
power, had battery backup. NDCC Operator #1 did not discuss these exact parameters of the
circuit, but based on SED’s understanding of the DMS system, the number of customers on the
circuit should have been obtainable.
126
The PI Historian database is the database that stores the SCADA data.
127
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
128
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
129
PG&E. “Document 1474-8 – 07 – 13 – 21 – 1125 – Dixie Troubleman – NDCC Operator #1,”
(Document 1474-8) (September 17, 2021), page 6.
130
United States District Court, Northern District of California, “Testimony of Troubleman” (September
13, 2021) page 50.
131
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
132
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
133
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
134
PG&E. “Paradise 21-0089207 Outages” (August 7, 2021), page 6.
135
Response to Data Request Dixie Fire-SED-005, Question 8, page 2.
136
See email message from Shawn Lanka, to Emily Fisher, CPUC Legal
Division, subject: “California Public Utilities Commission - Wildfire Safety, Request for
Information” (August 19, 2022). SED confirmed with the that it previously sent PG&E the
information confirming battery backups, although could not locate a record of the communication
to PG&E. SED also confirmed that in that location is not a critical customer as defined by
Decision 20-05-051, as the in that area is not used for civilian or military purposes.
Page 38 of 50
The Dixie Troubleman and NDCC Operator #1 spoke for the first time at 1125 hours on July 13,
2021.137 Their discussion indicates an extensive knowledge of the circuit and understanding that
access to Fuse 17733 would be time-consuming.138 During the call, the Dixie Troubleman
referenced the difficulty of accessing Fuse 17733. 139 NDCC Operator #1 stated that because the
load on the circuit appeared normally, the outage was probably related to the fuse. 140 NDCC
Operator #1 also referenced a permanent tag on the DMS indicating no access to Fuse 17733.141
No discussion of disconnecting power at Switch 941 as an alternative to accessing Fuse 17733
occurred.
After Hydro Operator #1 received an outage notification, they dispatched their Rover, who
142
confirmed there was an outage at Cresta Dam and the The Distribution Operator
requested PG&E Dispatch to assign a troubleman to investigate the outage further.143 PG&E
assigned the outage a Priority 1 tag, which was the correct Priority tag under relevant PG&E
procedures since the cause of the outage was unknown. A Priority 1 tag necessitates a same day
response, so a troubleman was dispatched.144
PG&E policy states that personnel can only turn off the power due to known hazards.145 Per
PG&E’s testimony, fuse operation does not constitute a known hazard.146 Accordingly,
preemptive de-energization of the line before the tree was known to be in contact with the line
would have been contrary to PG&E policy. However, de-energizing the line in this case would
have had minimal to no impact on customers, since two of the three customers were already
experiencing an outage and the third, the had battery backup.
137
Document 1474-8, page 1.
138
Document 1474-8, pages 4-6. The Dixie Troubleman and NDCC Operator #1 discussed the outage at
Cresta Dam and properties of Bucks Creek 1101 circuit, including the layout of the protection equipment
of Fuse 17733, Fuse 805 and Switch 941.
139
Document 1474-8, page 5.
140
Document 1474-8, page 6.
141
Document 1474-8, page 6.
142
20-Day report, page 2.
143
Document 1474-5, page 2
144
Response to Data Request Dixie Fire-SED-004, Question 53, page 1.
145
Response to Data Request Dixie Fire-SED-001, Question 8, page 1.
146
Document 1532-1, page 2.
Page 39 of 50
The Distribution Operators were also in charge of reviewing the load on the circuit. The
operators stated that they did not find any issues with the load.147 SED reviewed the load data
and found this conclusion to be generally correct. An analysis of the load data is discussed in the
Technical Addendum.
The Dixie Troubleman received the tag for the Cresta Dam outage at 1047 hours while enroute to
another priority one tag.148 After contacting the NDCC Operator #1 to learn more about the tag
related to the Cresta Dam outage, the Dixie Troubleman addressed the first tag.149 The Dixie
Troubleman arrived at Cresta Dam at approximately 1230 hours, approximately two hours after
receiving the tag and six hours after the original SCADA alert.150 He determined that Fuse 805
was not blown. The Dixie Troubleman continued to investigate the outage, walking to the
transformer located one span away on the line that runs to the Cresta Dam. The meter on the
transformer was off, indicating that the power was out. He continued to patrol the area to
discover the cause of the power outage. Using binoculars, he inspected the section of Bucks
Creek 1101 circuit that connect from the dam to Fuse 17733.151 Per the Dixie Troubleman’s
testimony in the District Court Case,
It appeared to me that all poles and wires on the line were up and in their normal
positions, not bent or twisted. However, I could see what appeared to be a fuse hanging
down from a pole on the circuit. That indicated to me that the fuse may have tripped on at
least one of the three phases of the line. At that point, I did not see any vegetation on the
line, nor did I see any smoke or other indication of fire.152
Figure 22 (see page 21 above) shows the span between Pole 100403908 and Pole 100403909 and
was taken after the Dixie Fire in 2021. In response to Data Request 4, Question 32, PG&E
provided a picture of the span taken in 2014 in advance of a pole replacement lower down the
hill (Figure 29, below).153 Though the primary focus of the photo in Figure 29 is not the span
near the fire origin, the photo shows the entire span near the fire origin.
147
Document 1515, page 15.
148
PG&E. “Document 1408-1 – Declaration of in Response to Order Requesting Information on
Dixie and Bader Fires” (July 28, 2021) (Document 1408-1), page 2. (Name of Declarant redacted in
original.)
149
Document 1408-1, page 2.
150
Document 1408-1, page 2.
151
Document 1408-1, page 3.
152
Document 1408-1, page 3.
153
PG&E Response to Data Request Dixie Fire-SED-004, Question 32 (photo attached as “PGE-DIXIE-
CPUC-000005204_CONFIDENTIAL”).
Page 40 of 50
before the bridge.156 The latter route appears to have taken 10 minutes longer than the first route.
Despite a road closed sign at the bridge, the Dixie Troubleman was able to cross.157 He arrived at
Pole 100403908 at 1640 hours and saw that two of the three fuses on the pole had operated. As
he exited his truck, he smelled smoke, but assumed it was coming from the Sugar Fire.158 He
opened the third fuse to prevent single phasing. From his vantage in the bucket lift, the Dixie
Troubleman “could see a fire downhill from [his] position,” further stating that the fire:
[was] about two-thirds of the way to the next pole. The fire was to the left side of the
right of way and roughly 600 or 800 square feet in an oval shape. The near edge of the
fire was not at the right of way; the far edge was roughly 25 yards from the right of way.
I could also see a tree leaning against the line. I did not see any breaks in the lines or
damage to other equipment. 159
The Dixie Troubleman radioed for help at 1655 hours after opening the third fuse.160 A
supervisor responded at 1656 hours.161 The Dixie Troubleman did not hear a response from the
supervisor and attempted to fight the fire.162 He emptied his extinguisher and returned to his
truck. He radioed again at 1706 hours and a supervisor responded on the radio at 1710 hours.163
After speaking with his supervisor, he returned to fight the fire.164 At this point he estimated the
fire was about 1200 square feet.165 At 1730 hours, CAL FIRE aerial assets arrived and started to
suppress the fire. At 1900 hours, a CAL FIRE ground crew arrived. Around this time, a PG&E
transmission supervisor arrived, and the Dixie Troubleman brought him to the site of the fire.166
After they returned to the bridge, a CAL FIRE investigator arrived. The Dixie Troubleman
informed the investigator of the two open fuses and the tree on the line.167 The two PG&E
employees left the scene at approximately 2000 hours.168
156
Exhibit X-2, page 2.
157
Document 1408-1, page 4.
158
The Sugar Fire burned northeast of Beckwourth, CA. Beckwourth is approximately 50 miles east of
Cresta Dam.
159
Document 1408-1, Page 4. The Dixie Troubleman’s description of the Dixie Fire when he arrived is
shown diagrammatically in Figure 3 on page 7.
160
PG&E “Document 1474-15 – Rough Transcript of July 13, 221 Dixie Troubleman Radio Calls
between 16:55 and 17:16” (September 24, 2021) (Document 1474-15), page 1.
161
Document 1474-15, page 1.
162
Document 1408-1, page 4.
163
Document 1474-15, page 1.
164
Document 1408-1, page 4.
165
Document 1408-1, page 4.
166
Document 1408-1, page 5.
167
Document 1408-1, page 5.
168
Document 1408-1, page 5.
Page 42 of 50
6. Overall Risk of Bucks Creek 1101 Circuit
SED investigated the overall risk of the Bucks Creek 1101 Circuit to contextualize PG&E’s
response to the initial SCADA alert at 0648 and subsequent outage at Cresta Dam. SED finds
PG&E in violation of PU Code Section 451 for failing to provide utility service necessary to
promote the safety of the public in failing to prioritize the response to the outage at Cresta Dam.
The SCADA alert occurred at 0648 hours, but the Dixie Troubleman did not arrive at Fuse
17733 until 1650 hours.173 NDCC Operator #1 was aware of the outage by 0721 hours after a
PG&E Hydroelectric Operator called to confer about the outage.174 The PG&E Hydroelectric
Operator sent a Rover to investigate the outage.175 The Rover verified the outage and that the
also lost power.176 The Dixie Troubleman was assigned to investigate further
and arrived at Cresta Dam at 1218 hours after addressing another tag.177 The Dixie Troubleman
arrived at the bridge at 1327 hours en route to look at Fuse 17733 and could not proceed at that
time due to bridge work.178 The Dixie Troubleman left and was unable to return until 1630 hours
due to the length of the road, even though the bridge work was done at approximately 1520
hours.179 The Dixie Troubleman did not arrive at the Fuse 17333 until 1650 hours, approximately
10 hours after the initial SCADA alert.
The bridge work impeding access to Fuse 17333 started between 0900 and 0930 hours.180 A
prompt, prioritized response from PG&E to the Cresta Dam outage could have enabled the Dixie
Troubleman to access the fuses before the bridge work started, or to stay at the bridge to access
Fuse 17733 as soon as the bridge was passable., or alternatively, the Dixie Troubleman could
have opened Switch 941 after returning to cell service on Highway 70 and realizing that
accessing Fuse 17733 would take additional time.
PG&E created the 2021 Wildfire Distribution Risk Model (Model) to rank circuit segments in
HFTDs based on wildfire risk.181 The Model is described at length in its 2021 Wildfire
173
20-Day report, page 2.
174
20-Day report, page 2.
175
20-Day report, page 2.
176
20-Day report, page 2.
177
20-Day report, page 2.
178
20-Day report, page 3.
179
20-Day report, page 3; CAL FIRE Investigation Report, page 37.
180
CAL FIRE Investigation Report, Page 37
181
PG&E Response to Data Request Dixie Fire-SED-004, Question 49 (October 8, 2021), page 1.
Page 44 of 50
Mitigation Plan. The Model is used to prioritize wildfire mitigation programs such as System
Hardening and Enhanced Vegetation Management.182 The Model is not used to inform decision-
making for purposes of operating the system.183 The Model has two components, the Equipment
Risk Model and Vegetation Risk Model.184 The Equipment Risk Model is the probability of
ignition from equipment failure, whereas the Vegetation Risk Model is the probability of ignition
from contact from vegetation.185 The Equipment Risk Model includes the chance of probability
from contact from vegetation.186 Bucks Creek 1101 was ranked 11 out of 3635 circuits for the
Equipment Risk Model and 568 out of 3074 circuits for the Vegetation Risk Model.187
The CAL FIRE Report noted the following regarding the risk of the area adjacent to Bucks
Creek 1101:
It is common and historic knowledge that the Highway 70 corridor is known for extreme fire
danger and poor access. Several large and devastating fires including the Camp Fire, (a
PG&E caused fire) have ignited over the last several years in that geographical area. It is also
common knowledge that the month of July in Butte County and surrounding areas is peak
fire season, yet no sense of urgency was demonstrated by PG&E to determine the cause of
the fault in a fire prone are during a severe time of year.188
Had PG&E responded to the alarm in a reasonably prompt manner, considering the extreme fire
danger, poor access, and history of previous wildfires, the Dixie Fire could have been prevented.
PG&E missed two potential opportunities to respond in time to prevent the fire. First, PG&E had
a 2.5-hour period in which to respond to the alarm and access the fuses before the bridge work
obstructed access. Second, the bridge work finished at 1520 hours, but the Dixie Troubleman
did not return to the bridge until 1630 hours.189 Assigning higher priority to trouble on the Bucks
Creek 1101 circuit, based on the well-established high wildfire risk associated with the circuit,
would have likely resulted in access to the fuses as soon as the road was passable, or other action
(e.g., de-energizing the circuit) consistent with a higher priority response.
182
Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
183
Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
184
Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
185
Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
186
PG&E 2021 Revised Wildfire Mitigation Plan – Revised (June 3, 2021), page 161.
187
Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
188
CAL FIRE Investigation Report, page 45.
189
20-Day report, Page 3.
Page 45 of 50
out of 3535 circuits in its Equipment Risk Model for wildfire risk.190 While SED understands
that this model is not used to inform operations decision-making, the model suggests that the
Bucks Creek 1101 is in the top one percent of the most hazardous circuits for wildfires. PG&E’s
response to outages on this circuit should account for this risk. The circuit is also located in a
Tier 2 HFTD. Further, as noted by CAL FIRE, the Highway 70 corridor (where the Dixie Fire
started) is an area of extreme fire danger and July was fire season.191
PG&E’s procedures for responding to outages or other types of trouble on a circuit do not
account for known risks. As a result, PG&E personnel failed to prioritize the response to the
alarm effectively and appropriately given the extreme wildfire risk associated with the circuit,
missing two chances to respond to the Cresta Dam outage in time to eliminate the fire risk.
Therefore, SED finds PG&E in violation of PU Code Section 451 for failing to provide electric
service as necessary to promote public safety.
8. Bucks Creek 1101 Circuit - Analysis of Current Data and System Data
SED investigated and analyzed the current and system protection design of Bucks Creek 1101.
SED finds no violations of GO 95 or the PU Code related to this portion of SED’s investigation.
Refer to the Technical Addendum for more information and analysis. A summary of the
conclusions from the Technical Addendum follows:
Current data provided by PG&E points to the tree hitting the power lines, causing a
phase-to-phase fault at 0648 hours.
Fuses at Fuse 17733 operated before Recloser 1101/2, which cleared the phase-to-phase
fault by shutting power off to two out of three phases of the circuit beyond the fuses.
After the fuses operated, the tree was still in contact with the third conductor, Phase A
(far left conductor on Figure 5 on page 10), which remained energized.
The contact caused a high-impedance fault, which eventually started a fire along the tree,
as shown in Figure 3 on page Figure 3.
If the recloser settings were set more sensitively than the fuses, the recloser would have
triggered, turned off power to all three phases, which would have prevented the fire. At
the time of this fire, however, fuses were typically designed to be more sensitive than a
recloser, so this was not common practice.192
190
PG&E Response to Data Request Dixie Fire-SED-004, Question 49, page 1.
191
CAL FIRE Investigation Report, page 45.
192
Since the start of the Dixie Fire, PG&E has implemented Enhanced Powerline Safety Settings (EPSS),
which implements settings for reclosers that are more sensitive than fuses.
Page 46 of 50
SED investigated two additional items: a helicopter flight performed by PG&E to look for smoke
near Cresta Dam on July 13, 2021, and a report of a drone in the vicinity of the fire. No
violations of GO 95 were identified related to either item.
At approximately 1500 hours, a Senior Power Generation Inspector at the Bucks Creek
Powerhouse parking lot thought he smelled and saw smoke southeast of the powerhouse.193
Cresta Dam and the Incident Location are roughly southwest of Bucks Creek Powerhouse.194 The
inspector called the Rock Creek Switching Center to alert the Operator that he smelled smoke.195
The Operator recruited a helicopter to fly over the area around Bucks Creek 1101.196 The
Operator also alerted an electrician at the Rock Creek Switching Center, who circled the facility
and did not smell smoke or see any sign of fire.197 The helicopter performed a fly-over of the
area.198 The pilot reported that he did not see a fire or any indication of a fire.199 The flight path
passed over a portion of Bucks Creek 1101 circuit, but the fly-over was focused near the Bucks
Creek Powerhouse and Rock Creek Powerhouse (Figure 30).200 The closest point to Pole
100403909 from the flight path was approximately one mile, but the flight path and Pole
100403909 were on opposite sides of a mountain ridge.201 PG&E is not aware if any of the
information related to the suspected fire was conveyed to a Distribution Operator at the NDCC
or the Dixie Troubleman.202
193
Document 1479, PG&E Further Response re Dixie Fire 092421 (Doc. 1479), page 10.
194
Doc. 1479, page 10.
195
Doc. 1479, page 10.
196
Doc. 1479, page 10.
197
Doc. 1479, page 10.
198
PG&E Response to Data Request Dixie Fire-SED-005, Question 18 (February 28, 2022), page 1.
199
Response to Data Request Dixie Fire-SED-005, Question 18, page 1.
200
PG&E “KMZ Map of Helicopter Flight” (PGE-DIXIE-CPUC-000017677) (July 13, 2021).
201
PGE-DIXIE-CPUC-000017677
202
PG&E Response to Data Request Dixie Fire-SED-005, Question 20 (February 28, 2022), page 1.
Page 47 of 50
Plumas County, but PG&E records indicate that the closest authorized drone flights were 20
miles away from the Incident Location.205
Drone Analysis
Based on evidence that the drone was not associated with PG&E operations, no violations of GO
95 are identified regarding a drone flight.
Conclusion
IV. Conclusion
A. Violations
Based on the evidence reviewed, SED’s investigation found six violations of GO 95 and one
violation of PU Code Section 451 by PG&E:
1. PG&E’s failure to complete Electric Overhead Tag 109671451 within the required one-
year deadline is a violation of GO 95, Rule 18.B.
2. PG&E’s failure to maintain records that show the correct date of inspection for its 2020
vegetation management routine inspection is a violation of GO 95, Rule 18.B.
3. PG&E’s failure to maintain a complete set of records from its 2019 vegetation
management routine inspection is a violation of GO 95, Rule 18.B.
4. PG&E’s failure to update Appendix A from its vegetation procedures to appropriately
reflect the MDRs required by GO 95, Rule 35 is violation of GO 95, Rule 31.1.
5. PG&E’s failure to identify the tree on the line which was reasonably visible from Cresta
Dam is a violation of GO 95, Rule 31.1.
6. PG&E’s failure to maintain its 12 kV overhead conductors safely and properly is a
violation of GO 95, Rule 31.1. PG&E did not identify a hazardous tree condition and
take the appropriate steps to prevent the Subject Tree from striking the overhead
conductors.
7. PG&E’s failure to adequately consider the hazard of Bucks Creek 1101 circuit in its
response to the outage at Cresta Dam is a violation of PU Code Section 451.
https://www.audacy.com/kcbsradio/news/weather/drone-interfered-with-critical-efforts-to-control-dixie-
fire (August 12, 2021 News Item).
205
PG&E. “Response to Data Request Dixie Fire-SED-001, Question 2,” Page 1. May 25, 2022; see also
August 12, 2021 News Item, quoted Butte County District Attorney Mike Ramsey as stating that the
drone was not likely to be PG&E’s.
Page 49 of 50
If SED becomes aware of additional information that could modify SED’s findings in this
Incident Investigation Report, SED may re-open the investigation; if so, SED may modify this
report and take further actions as appropriate.
Page 50 of 50