San Bruno Fire After Action Report/Technology Debrief

San Bruno Fire Blog Post

This is transcribed from the After Action Report from the unofficial Technology Debrief after the San Bruno Gas Explosion and Fire.  The Debrief was called by me (Luke Beckman), formerly the Director of Disaster Operations for InSTEDD, and 32 people participated in the debrief hosted at Google. The After Action report and its excerpts were compiled by Carnegie Mellon University-Silicon Valley’s (CMU-SV) Disaster Management Initiative (DMI) team. There was a wide range of participants, representing many critical organizations involved in disaster operations in the San Francisco Bay Area in California.

Case Study- San Bruno Gas Pipeline Explosion:

At 6:00 PM PDT, on Thursday Sept 9, 2010, there was a rupture in a two-foot gas pipeline in San Mateo County, California, escalating to a six-alarm fire by 7:23 PDT that ended up completely destroying 37 homes and damaging an additional 18. There were eight fatalities, dozens of injured, and hundreds were evacuated.  At 8:52 PM, a CAL-EMA Fire Management Assistance Grant was submitted. Strike, search and rescue, and damage assessment teams were all ready and staging by 9:00 PM, but would ultimately not go out until the scene was safer at first light Sept 10. At 10:25 PM, I sent out an email to emergency managers across the Bay Area asking “how can we coordinate effectively?”

This email triggered a complex set of coordination activities that interlinked formal and ad-hoc crisis response capabilities for the rest of the event. In response to the email from me, the Planning Section Chief on scene requested advanced mapping and imagery capabilities. Mapping personnel would not ultimately affect fire suppression operations, but rather allow certain members of the incident command team to have better situational awareness, and to allow fire investigators and search teams to have a more accurate picture of the scene prior to deployment.

The first team of three mappers was dispatched to the Incident Command Post, at no cost, at 1:12 AM Sept 10, and they arrived at 2:00 AM. All information that had been requested by the Planning Section Chief was ready by 7:00 AM for the Incident Planning Briefing. Shortly thereafter, aerial imagery had been taken by one of the mapping team, working in tandem with a CalFire Air Attack crew. Over the next several days, approximately six mapping teams would work in direct support of incident operations, exceeding the requests of the Incident Command staff.

This adhoc capability was only unleashed and leveraged because of close personal relationships because of the high level of flexibility exhibited by the volunteer mapping teams, and because of the swarm potential of the volunteer technical community. This is NOT standard operating procedure, and emergency managers cannot, today, reliably depend on this level of capability, working with this level of speed and flexibility—but they could be. We should expect and demand this level of efficacy and response across the country, from the neighborhood to the federal levels.

Menlo Park Fire Chief, Harold Schapelhouman, who was the original Planning Section Chief at the incident, wrote an email to me and a few other individuals a few days after the fire was suppressed. It read:

“I can’t begin to express our appreciation for what you all did and continue to do for this incident. Thank you for reaching out and allowing us to take advantage of not only our relationship but also the talents of your organization and network. I haven’t seen all of the good work that was done but I know you guys were considered super stars and I think in the days, weeks and months to come the quality of your work and capabilities of merging advanced mapping and information technology with first responders will be seen for the tremendous benefit that it was to this incident.”


Excerpts from the After Action Report: Some Observations and Lessons Learned

  • The initial volunteer teams, once assembled, were coordinated and dispatched using a GeoChat text message group (with zero previous training and the group was stood up on the fly). This was a highly effective means of communication, but the mapping component of GeoChat was not utilized. GeoChat is a free and open source, multi-channel group coordination tool that was developed by InSTEDD.
  • There was duplication of mapping effort, by several different actors including: relief teams lacking information or access to previous work; by teams on the ground (e.g., ESRI folks mapping in the Emergency Operations Center [EOC] and volunteer mappers at the Incident Command Post [ICP]); by different agencies (e.g., work conducted by a volunteer working with PG&E which overlapped with volunteer mapper work).
  • Even several weeks after the incident, all professional and volunteer activities are not known. For example there are stories still floating around that this group has not been able to check: San Francisco Fire Department saved a neighborhood from being burned by covering homes in fire retardant foam, many animals were rescued, burned victims were found by local citizens, etc…. On the law enforcement side, a lot happened as well, but these entities have not been involved in this specific process- it is heavy on the fire side.
  • The National Transportation Safety Board’s (NTSB) job improved with early GPS tags (provided by the mapping team) of fluorescent-tagged blast sites before scene was disrupted. This made data analysis possible; they did not have that level of detail before this specific incident.
  • Information was lost in hand-offs. For example: At the EOC, the point person’s email address was lost in the transfer of teams around 6:30am Saturday; URL location of google team “my maps” did not get transferred to next team. Reasons include: not knowing who to hand off information to; not giving information to the right person (e.g., The Google team gave login information to people working in their area, but it didn’t make it to the hand-off team). The first wave of volunteers transferred KML files to the Google team, but they did not receive GIS or paper maps.
  • Communication among professional groups initially was with face-to-face meetings. (Teams would initaially drive in a truck from the ICP to the EOC to show the Google team’s maps). The team eventually got the main email address for the EOCso that mappers didn’t have to drive to the EOC.
    • Missing communication links: San Bruno EOC feedback needed to field EOC – names of people needing transport. Need help in search phase. Need who is missing. ICP did not know that the CISCO Networked Emergency Response Vehicle (NERV) had been put on alert or even stood down.
    • San Mateo Sheriff Search and Rescue team didn’t know about availability of map products from mapping volunteers.
  • Mapping was done online and offline. Transition across online and offline was not always smooth.
  • Paper maps were initially made by driving to Kinko’s or by using printers in the EOC.
  • Cell phone connectivity was patchy and especially poor at night. Mapping teams established connectivity with a combination of phones of different carriers, and used a mobile WiFi hotspot.
  • Maps showing impacted area, total number of dwellings were priceless for American Red Cross for planning and locating evacuation and shelter sites. [Note – this statement was recorded from George Smith, but Mark Lowpensky notes: “I know the evac site was identified before any mapping was done and I believe the shelters were also all identified before any mapping was started.” (These items both occurred Thursday night and that statement has been corroborated)]
  • Missing data cross-linking. APN numbers and who owned what homes would have helped.
  • Volunteer mappers were deployed to the ICP and EOC.
  • How volunteers were recruited – tenuous connection, relied on personal relationships. The only way that volunteer mappers were recruited was because Harold Schapelhouman had worked with Luke Beckman and knew of his work in Haiti. Chief Schapelhouman was working with e-mail to send the FMAG (CAL-EMA Fire Management Assistance Grant email) when he saw e-mails from both Peter Ohtaki of the CA Resiliency Alliance and Luke Beckman offering assistance. Luke used his personal connections to activate volunteers in the field. Jeannie Stamberger, of the DMI, used Twitter and volunteer group to recruit mapping volunteers. Concern on the tenuousness of this connection (e.g., what if Luke was asleep?), and the lack of availability of volunteer resources to those without these personal connections. This needs to change to a system which has more connections (making it more robust) and where use of volunteers is not dependent on personal connections.
  • Could have provided more volunteer mappers. Two volunteer mappers were sent to the field on Saturday, based on known needs. However we received 7 offers of assistance within 2 hours of the notice going out, with 7 additional offers and an offer of a gigapan by noon on Saturday, and 4 additional offers of assistance Saturday afternoon including a trained survey mapping and 3-D mapping team (Terrain Lab from SRI) and a forensic anthropology team (Foothill college Dept of Anthropology. Because the request went out at 8pm on Friday night, offers of assistance were low because many people had logged off for the night. Because it was over a weekend we only received notice on Monday about the availability of the Nokia/Navtek 3-D laser image car; requests would have pursued availability if the need was expressed. Note from Mark Lowpensky of San Mateo SAR:  “I am not sure if it was the volunteered Gigapan, but NTSB did take panoramic pictures of the scene.”
  • Difficulty of Volunteers Offering services: Cisco knew the event happened immediately and were prepared to offer their services, however, they couldn’t contact the EOC because Cisco didn’t have cell phone numbers of people in the EOC. Harold noted phone numbers aren’t usually helpful because everyone is calling them. Solution: be part of a pre-packaged process incorporated into Computer Aided Dispatch so you get the call rather than calling them.  Mark Lowpensky notes re Cisco and the EOC: “It is better to call dispatch and use land lines and not rely on cell phones (being in CAD is the best, but you could always get a message to someone in the field through dispatch.”
  • Cisco observed ICS worked– fire and law worked well together. Harold Shapelhouman said they used a “border-drop” system that was very effective. This system essentially allows resources from one jurisdiction to seamlessly support efforts in another jurisdiction when summoned. Each area covers the back of other areas.
  • For the first time a public “my map” generated for the emergency was spammed and had to be shut down.
  • Data accuracy was a problem in the field. It would have been helpful to have phone numbers of people providing data or making previous iterations of maps to call when a data conflict arose.
  • Manual geo-tagging was conducted. Google team went out with cameras and GPS and pen and paper and recorded images of damaged cars (spray-painted with a number) and GPS waypoints of damaged cars and houses. This process was manually cross-referencing uploaded photos in picassa, and linking them with geotagged waypoints, (B for building). Data was not complete. Photos are very useful (worth a thousand words).  Geotagged photos are really helpful…When data were aggregated.. with paper mapped data, and data was incomplete and inaccurate. Orientation of image is important.
  • Mappers received data dumps from responders. People would come by and drop off paper – written on paper, hand drawn things. Responders didn’t know how to effectively interface with volunteers.
  • Primarily one way data transfer. While people gave data to be entered into maps, maps containing this information were not widely requested. This could be 1) people didn’t know about mapping efforts 2) didn’t know how to use
  • Observations from Mark Lowpensky
    • – Need for hardware for outputting maps
    • – Confusion caused by open vs secured wireless access points (concern of media using networks is one of the reasons for them to be closed)
    • – Volunteers need to come up with a list/cheat sheet of things we might ask the EOC for, such as to use ArcView to export the parcel data for a specific area so we don’t need to deal with data for the whole city/county, or access into ESRI. (Google Earth Pro may not choke, but it will take a LONG time.)
  • Aerial Imagery– The Incident Command Team wanted aerial imagery- they immediately thought super high tech and “Satellites.” Due to the lengthy process of getting the required permissions, retasking a satellite, possiblity of smoke or fog cover, and the short time window in which to shoot before an 8:00 AM briefing on Friday morning, the decision was made on around 5:00 AM to use a Cal Fire plane and a digital camera. These photos were then geocoded on Google Earth and delivered to the State, federal entities, and the Governor’s office.

Background for Technical Volunteers/Responders

  • There are two types of maps: strategic and tactical. Strategic maps may be produced every 12 hours and will be used to provide a situation summary to city managers, media etc. Tactical maps are small-scale maps specific for a task/team and are produced every few hours. Tactical maps are printed maps used in the field with constantly updated information.
  • Professional first responders want to use paper maps. Getting color maps in the field is still an issue; these are often needed. Laser printers are preferred to inkjet, because ink from inkjet printers run when paper is wet. Additional mapping features may improve utility of printed tactical maps such as shrinking placeholder, universal utility in black/white or color printing, and optimizing for small-scale maps (e.g., the scale of a few houses).
  • Not having to pay for civilian services speeds use of services. Payment requires figuring out who will pay, delaying use of services. Or if things were prepaid or covered…
  • What is an ICS Information Unit? The content of an “Information Unit” depends on if it is Fire or Law. It can be for public information or investigation purposes (e.g. curates clues in an investigation for detectives); however, it is not general information support for emergency ops, such as maps, computers, IT, communications. The latter is an unmet service by the ICS that we could create and civilians provide.
  • When generating maps, must be careful about the level of detail which is made publically available, or put on strategic maps shown to the media. For example, maps made presented to media removed location of fatalities. However, this information is extremely important on tactical maps.
  • ICS does not have a post for interacting with civilians. The ICS post who knows who has pre-existing intel is probably logistics.
  • Incident Action Plan (IAP) is normally updated for each operational period (8 or 12 hours – according to Mark Lowpensky),  and printed for the next period (although Jim Varner notes they are produced every few hours). These are needed as hard copies (printed on paper) and digital copies (because most emergency operators are using computers (Word Docs or PDF forms to fill in) to generate the IAPs). Integration of maps into IAPs is needed: For example, South san Francisco chief who asked for printouts to embed into IAP.
  • Difficult to offer services currently. Even if you have the phone number of a commander, they receive so many calls you will be unlikely to get a hold of them. Cisco had the problem of not having phone numbers to call when they heard about the event.
  • How to get through. It is better to call dispatch and use land lines and not rely on cell phones. Being in  CAD is the best, but you could always get a message to someone in the field through dispatch. (-Mark Lowpensky). Pre-event training and familiarization can work to solve this problem.

Comprehensive Timeline of Events

September 9, 2010 (Thursday)

  • 18:00
    • 18:11 – USGS Richter Scale 1.1 event due to explosion
    • 18:12 – Initial dispatch SMC Public Safety Communications Center
    • 18:14 – Second Alarm
    • 18:19 – Third Alarm
    • 18:22 – Fourth Alarm
    • 18:25 – Alameda County Strike Team Request
    • 18:33 – Fifth Alarm
    • 18:44 – Chief 1 – Attached to Incident
    • 18:46 – Alameda County Strike Team Request
  • 19:00
    • 19:23 – Sixth Alarm
  • 20:00
    • 20:52 – CAL-EMA Fire Management Assistance Grant (FMAG) Submittal
  • 21:00
    • CISCO NRV requested Thursday night by CalFire around 9pm/10pm to provide comm. (later stood down)
    • Chief George Devendorf teams are ready to do a damage assessment at 9pm, but could not go out until the morning because of safety.
    • UNKNOWN TIME: From Mark Lowpensky- I know the evac site was identified before any mapping was done and I believe the shelters were also all identified before any mapping was started. (These items both occurred Thursday night)
  • 22:00
    • 22:25 – Luke Beckman’s email – San Bruno Fire – How can we coordinate effectively?
  • 23:00
    • over 6 hours later at the command level people don’t really know the situation. This is not unusual, because early stages of disasters are  dynamic situations. For example, the pipeline was open for 2 hours,  People dead, etc

September 10, 2010 (Friday)

  • 00:00
  • 01:00
    • 0112 – Luke email – Names of Mapping team (Clay Zach and Jake)
  • 02:00
    • Arrival of time of mapping team briefing (estimate) and 3 initial objectives.
    • mapping team collects at 2am, give maps to everyone at the command post, but not good connectivity from ad hoc field set up to the EOC. Zach and Harold drove and met and given the map – questioned level of accuracy.
    • 02:30am City manager had sig/mod/light assessment of damage to homes, rough number of cars, burn victims, fatalities.
    • 02:38:  Luke Beckman calls Jeannie Stamberger to NASA UAVs with IR capability to see through fog (J. did not receive call; UAVs unlikely to have been permissible)
    • 02:45: meeting with the San Bruno EOC email address – Request Fatality Layer.
    • Need map for 7/8am public information meeting. Concerned with use of map because it showed exact location of the homes – mapping team eliminated that information.
  • 03:00
  • 04:00
  • 05:00
  • 06:00
    • 0630- Deputy Chief Campbell (MPFPD) arrives and assumes planning activities
  • 07:00
    • 0700 – Incident Planning Briefing
  • 08:00
    • Search, Investigation and Damage Assessments (Information Requests)
  • 09:00
    • Clay Sader begins process of acquiring aerial imagery in a CalFire plane.
  • 10:00
  • 11:00
    • Google mapping group is on the scene (Brian and Christiaan); receive KML files from first wave team (Zach Jake and Clay)
  • 12:00
  • 13:00
  • 14:00
    • 1400 Site Walk with Brian and Christian – GPS layer home and Vehicle Layer
  • 15:00
    • Jeannie calls Luke to see if he needs any assistance.
    • 1530 Christian asked if we wanted the pan-cam  – confirmed we wanted it.
  • 16:00
  • 17:00
  • 18:00
    • Luke requests mapping volunteers via phone.
  • 19:00
    • Jeannie starts recruiting mapping volunteers using Crisiscamp volunteers registrants, crisiscommons google group lists, dmi-group email list, dmi LinkedIn account and cmusv-dmi Twitter account, and webmapsocial group.
    • 19:48 Mapping volunteer emails and Eventbrite requesting assistance created  “URGENT – San Bruno Mapping Posse”; send to 250 members attending CrisisCampSiliconValley. Tweets also sent out.
    • 19:56 First response to request
  • 20:00
    • 20:00: San Mateo Search and Rescue found out about mapping , but there was no way to create hard copy maps.
    • 20:00 Volunteer mapping Email sent to CrisisCommons
    • 20:12: First retweet of volunteer request. RT @cmusv_dmi: URGENT: Officials need expert volunteer mappers TOMORROW (Sat am) on ground to map San Bruno fire #sanbrunofire…
    • 20:46: DMI members active on tweeting San Bruno fire info: RT @cmusv_dmi: Twitter list for San Bruno fire updates #sanbrunofire #fire #sanbrunofire
  • 21:00
  • 22:00
    • Google mappers had created non-public Google earth “My maps” with agency information and were prepared to hand off the information, but didn’t know how/who to hand off to.
    • By this time we have 4 volunteers and send out message that the need is filled. First retweet of that message at 22:03: RT @cmusv_dmi: NEED FILLED – Thanks! re: expert volunteer mappers TOMORROW … San Bruno fire #sanbrunofire
    • By 22:17 we have 5 qualified tech volunteers recruited to map at 8am Saturday morning
  • 23:00

September 11, 2010 (Saturday)

  • 00:00
  • 01:00
  • 02:00
  • 03:00
  • 04:00
  • 05:00
  • 06:00
  • 07:00
  • 08:00
    • Volunteer mappers Sumathi and Sanjay arrive at San Bruno EOC for duty
    • 8:20 AM    Lessly Fields    Call from PG&E for help.  Coordinate utilities and mapping. Lessly would get me mapping people contacts
    • 8:21 AM    Coggeshall    email First Map complete and up on
    • 8:42 AM    Lessly Fields    email with location information
    • 8:53 AM    Riordan    call to get local water utility
    • 8:55 AM    Leles    call to get local water utility
    • 8:57 AM    Eric Beckles    call to get local water utility
  • 09:00
    • 9:10 AM    Lessly Fields    email requesting CUEA support
    • 9:11 AM    Henry Degroot    call to get local water utility
    • 9:13 AM    Simunovich    call to get local water utility
    • 9:18 AM    Coggeshall    Email about Google request for mapping
    • 9:19 AM    Coggeshall    Call to get mapping started
    • 9:24 AM    Simunovich    email with cal water contacts
    • 9:25 AM    South City    Call Water
    • 9:29 AM    Lessly Fields    email with LAC and Logistics contacts
  • 10:00
    • 10:03 AM    Lessly Fields    email with phone number for logistics
    • 10:08 AM    Lessly Fields    Give update and get logistics person
    • 10:16 AM    Lessly Fields    Email connecting me to PG&E GIS
    • 10:16 AM    Tony Carasco    Email from Cal water connecting me to water utility
    • 10:16 AM    Elizabeth Proctor    Email from GIS starting mapping and laptop
    • 10:17 AM    Elizabeth Proctor    Start coordination for mapping, put Elizabeth in contact with Dave
    • 10:17 AM    PG&E E&O    Contact logistics for status on Laptop for LAC
    • 10:24 AM    South San Francisco 2    Contacted Water Utility for San Bruno
    • 10:28 AM    Eric Beckles    Eric gave me his contacts for water company
    • 10:30 AM    South San Francisco    Talked to Public Works
    • 10:35 AM    Doug Wisman, CalEMA    Doug put me in touch with Public Works rep at the LAC
    • 10:54 AM    Coggeshall    Update from Dave on Mapping Efforts
  • 11:00
    • Sumathi and Sanjay receive instructions
    • 11:05 AM    Elizabeth Proctor    udate on mapping and delivery of the laptop
    • 11:10 AM    Coggeshall    Update on status
    • 11:16 AM    Lessly Fields    Gave lessly and update on status of utilities and mapping
    • 11:30 AM    Coggeshall    Email Dave starting mapping with PG&E
    • 11:40 AM    Lessly Fields    Followup on laptop and status at LAC
    • 11:50 AM    Elizabeth Proctor    email data for first map coming soon
  • 12:00
    • 12:05 PM    Elizabeth Proctor    Update on laptop delivery to LAC and hand off to staff
    • 12:18 PM    Steve Frew    Advised that he would not be needed to take over the LAC utilities desk
    • 12:19 PM    Elizabeth Proctor    Email data sent
    • 12:55 PM    Coggeshall    Email 65 Meg’s of data to dave
  • 13:00
    • 1:36 PM    Coggeshall    Update on Dave working with data
  • 14:00
    • Hole was pumped out at 2pm in afternoon on Saturday .
    • 2:19 PM    Henry Degroot    Followup on earlier call
  • 15:00
    • 3:47 PM    Elizabeth Proctor    mapping update
    • 3:53 PM    Elizabeth Proctor    mapping update
    • 3:54 PM    Elizabeth Proctor    Email Jody from PG&E Google Earth joins group
    • 3:54 PM    Coggeshall    Email – Dave connected with Jody
  • 16:00
    • 4:01 PM    Cummings    email jody will have kml file soon
    • 4:14 PM    Coggeshall    email, dave received first data and producing map
    • 4:35 PM    Coggeshall    Mapping update
    • 4:52 PM    Cummings    email Jody prepared KMZ file for google earth
  • 17:00
    • Sumathi and Sanjay leave EOC
    • 5:57 PM    Coggeshall    Google Earth Update
    • 5:58 PM    Coggeshall    email update from dave
  • 18:00
    • 6:31 PM    Cummings    emailed updated kmz file for google earth
    • 6:41 PM    Coggeshall    email new website up with multiview
    • 6:59 PM    Coggeshall    Map moving to Google Earth
  • 19:00
    • 7:00 PM    Wollbrinck    Email to lessly to password protect map
    • 7:01 PM    Lessly Fields    email, agreed with PW and Thanked for help
  • 20:00
  • 21:00
  • 22:00
  • 23:00

September 12, 2010 (Sunday)

  • 12:11 PM    Coggeshall    Mapping update
  • 12:34 PM    Coggeshall    Mapping update
  • 2:55 PM    Coggeshall    Mapping update
  • 4:58 PM    Coggeshall    Map now on web, my maps and google earth

Next Steps resulting from the event and the Debrief

A working group was established, with meetings and sustained interest by the Carnegie Mellon Silicon Valley Disaster Management Initiative (DMI) to develop solutions within the geographic area of San Mateo County (particularly Menlo Park and San Bruno) initially focusing on mapping data/technology to communicate among people and the data infrastructure move information, involving the following actors: NGOs/for profit/for profit tech/volunteers/Emergency management/government, and organization that have overlaps among them. Projects for immediate development include: 1) Push Package of Computer-Aided Dispatch (CAD)-dispatched pre-identified resources 2) Registry of Free Volunteer Civilian Services, 3) Emergency Mapping Volunteer Operating Manual 4) Tool Development for Emergency Mapping Needs.

Summary of Immediate Projects: 1) Push Package of CAD-dispatched pre-identified resources 2) Registry of Free Volunteer Civilian Services, 3) Emergency Mapping Volunteer Operating Manual 4) Tool Development for Emergency Mapping Needs. Detailed descriptions are below.

Push Package of CAD-dispatched pre-identified resources
A push package would be a CAD-activated package of pre-identified resources identified in advance of the event which can be called upon by professional first responders during an emergency. This would avoid the problem of requiring personal relationships to call on civilian resources and the problem of someone not getting the call.  Initially identified components included:

  • Network of people to call on
      1. NetGuard FEMA pilot project to org IT, vol, equip, to provide temp. IT service to EOCs or relief organizations; stage is identifying potential needs (e.g., mapping)
      2. Disaster Asset Registry (AidMatrix, contact is Keith Thode). Identify private sector donated resources that could be donated under incident command with emergency contact info of people in the company who can release that resource (e.g., CISCO NRV); targeted around donated resources; stage:  populating registry.
    • InfraGard Bank of volunteers who have been screened through FBI records check. InfraGard and NetGuard are making some effort to coordinate with each other. Questions can be directed to Rich Davies of the Western Disaster Center, and a CMU DMI affiliate.
  • Access to standard formats for emergency mapping (either through use by civilian mappers or first responders)
  • Process is needed wherein emergency responders can work in advance with a resource representative to establish which packages they are suited to and then to have them added to the dispatch package.

Harold Schapelhouman notes that they don’t have a package for a field command post now.
The working group would work to develop additional components for the push package based on needs of professional first responders at the San Bruno fire.

Registry of Free Volunteer Civilian Services
The registry provides a network of civilians skills (e.g., technical skills such as computer mapping, veterinarians) for emergency responders to immediately draw on in an emergency. (The registry may focus on individuals contributing their time rather than organizations) The registry would be incorporated into the computer aided dispatch system, so professional first responders can access civilian resources without  personal connections. The registry would have lists of phone numbers and skills/talents/capabilities of civilian volunteers offering their services for free in an emergency. The registrants would be:

  • credentialed,
  • be managed with a civilian command and control system,
  • have a code of conduct,
  • be briefly familiar with ICS and the rigors of working in a ICP
  • show up supplies to support themselves for 12 hours
  • show up with basic equipment to accomplish their missions (e.g., laptops)
  • focus on mapping, but ultimately include a range of skills requested by EOCs (e.g. veterinarians to support SAR activity)

A prototype registry will be developed immediately using list of people attending the debrief meeting and volunteer mappers that responded to the call (see Appendix “Civilian Resources Network”). Maturation of the registry will require addressing several issues which need to be resolved to achieve a working registry:

  • How to credential volunteers
    • About: Problem with the friend of a friend is they can’t cross the security line. Harold had to get these guys to get across the line. Security gets tighter,  no uniform or ID.

Emergency Mapping Volunteer Operating Manual

  • 1-2 pager “operating manual” for mapping volunteers. This can address several issues:
    • Credentialing;
    • List of standard operating procedure
    • Smooth transition across shift teams; hand-off; to avoid loss of information
    • battalion chief is the intersection…
    • avoid duplication of work,
    • ensure mapping people are connected with the right people on scene and understand different places where mapping may be going on, the different types of information.

Tool Development for Emergency Mapping Needs

  • Custom icons in a tray. Use Google in-house disaster items as starter list?
  • Feature requests for mapping tools – e.g., being able to print at scale, gridlines for UTM and NGS grids, mechanism for tiling
  • Create standards and best practices for emergency mapping volunteers. Standards would include data formats (e.g., KML, .csv).  Standards/best practices would meet a number of needs identified in San Bruno experience including:
    • smoothing the online/offline transition
    • prevent need for duplication
  • Immediate mapping need: Ask Harold Schapelhouman if he needs technical assistance with final mapping layers he identified
  • Development of a standard push package or template for emergency mapping –e.g., it would have necessary layers already included, appropriate icons, data in standard formats etc. The working group will identify what this should look like.

NTSB Official Report

National Transportation Safety Board

Washington, D.C. 20594

Preliminary Report

  • Accident No.: DCA10MP008
  • Type of System: 30-inch natural gas transmission pipeline
  • Accident Type: Pipeline rupture
  • Location: San Bruno, CA
  • Date: September 9, 2010
  • Time: About 6:11 p.m., Pacific Daylight Time
  • Owner/Operator: Pacific Gas & Electric Company
  • Fatalities/Injuries: Eight fatalities, multiple injuries
  • Pipeline Pressure: 386 pounds per square inch gauge (psig) at the time of rupture
  • Quantity Released: Approximately 47.6 million standard cubic feet (MMSCF)

On September 9, 2010, at approximately 6:11 p.m. Pacific Daylight Time(1), a 30-inch diameter natural gas transmission pipeline (Line 132) owned and operated by Pacific Gas & Electric Company (PG&E) ruptured in a residential area in San Bruno, California. On September 10, the NTSB launched a team to California to investigate this tragedy. Vice Chairman Christopher Hart was the NTSB Board Member on scene in San Bruno.

The rupture on Line 132 occurred near mile post (MP) 39.33, at the intersection of Earl Avenue and Glenview Drive in the city of San Bruno. Approximately 47.6 million standard cubic feet (MMSCF) of natural gas was released as a result of the rupture. The rupture created a crater approximately 72 feet long by 26 feet wide. A pipe segment approximately 28 feet long was found about 100 feet away from the crater.  The released natural gas was ignited sometime after the rupture; the resulting fire destroyed 37 homes and damaged 18. Eight people were killed, numerous individuals were injured, and many more were evacuated from the area.

The Incident Command was set up by the local fire department. The immediate response by local emergency responders, as well as three strategic drops of fire retardant and water by air, assisted in stopping the spread of the fire.

According to PG&E records, Line 132, which is regulated by the California Public Utilities Commission (CPUC), was constructed using 30-inch diameter steel pipe (API 5L Grade X42) with 0.375-inch thick wall. The pipeline was coated with hot applied asphalt, and was cathodically protected. The ruptured pipeline segment was installed circa 1956. The specified maximum operating pressure (MOP) for the ruptured pipeline was 375 pounds per square inch gauge (psig). According to PG&E, the maximum allowable operating pressure for the line was 400 psig.

Just before the accident, PG&E was working on their uninterruptable power supply (UPS) system at Milpitas Terminal, which is located about 39.33 miles southeast of the accident site. During the course of this work, the power supply from the UPS system to the supervisory control and data acquisition (SCADA) system malfunctioned so that instead of supplying a predetermined output of 24 volts of direct current (VDC), the UPS system supplied approximately 7 VDC or less to the SCADA system. Because of this anomaly, the electronic signal to the regulating valve for Line 132 was lost. The loss of the electrical signal resulted in the regulating valve moving from partially open to the full open position as designed. The pressure then increased to 386 psig.  The over-protection valve, which was pneumatically activated and did not require electronic input, maintained the pressure at 386 psig.

At about 5:45 p.m., the SCADA system indicated that the pressure at Martin Station, which is downstream of the rupture location, exceeded 375 psig. The SCADA system indicated that the pressure at Martin Station continued to increase until it reached about 390 psig at about 6:00 p.m. At 6:08 p.m., it dropped to 386 psig. At 6:11 p.m., the pressure at Martin Station decreased from 386 to 361.4 psig; within one minute the pressure dropped to 289.9 psig.

PG&E dispatched a crew at 6:45 p.m. to isolate the ruptured pipe section by closing the nearest mainline valves. The upstream valve (MP 38.49) was closed at about 7:20 p.m. and the downstream valve at Healy Station (MP 40.05) was closed at about 7:40 p.m. Once the ruptured section was isolated and the gas flow was stopped, the resulting fire from the ruptured line self-extinguished. Later that evening, PG&E isolated the natural gas distribution system serving residences in the area, and within a minute of stopping the gas flow at about 11:30 p.m., fires from escaping natural gas at damaged houses went out.

When the NTSB arrived on scene on September 10, the investigation began with a visual examination of the pipe and the surrounding area. The investigators measured, photographed, and secured the approximately 28-foot-long ruptured pipe segment. On Monday, September 13, the ruptured pipe segment and two shorter segments of pipe, cut from the north and south sides of the rupture, were crated for transport to an NTSB facility in Ashburn, Va., for examination.

The examination revealed that the ruptured segment was 27 feet 8 inches long at its longest length, and consisted of a pipe section and four smaller pipe pieces (pups) between 3 feet 8.5 inches and 3 feet 11 inches long (pups are numbered one through four from south to north).

The segment north of the rupture (north segment) was 15 feet 9 inches long and consisted of a pipe section and two pups, 3 feet 7 inches and 4 feet 7 inches long (numbered five and six from south to north).

The section south of the rupture (south segment) was 12 feet 4.5 inches long at its longest length; it contained no pups.

All pipe pieces and pups showed fairly uniform wall thickness of 0.36 to 0.38 inches.

There were longitudinal fractures in the first and second pup of the ruptured segment and a partial circumferential fracture at the girth weld between the first and second pup. There was a complete circumferential fracture at the girth weld between the fourth pup in the ruptured segment and the fifth pup in the north segment. The longitudinal fracture in the first pup continued south into the pipe ending in a circumferential fracture in the middle of the pipe.

The following laboratory work on the pipe has been completed:

  • Written documentation, photo documentation and visual inspection of the pipe.
  • Removal of the asphalt coating from outside of the three pipe segments in preparation for non-destructive examination work.
  • Radiography of the girth welds and select seams.
  • Microbiological testing of the pipe surface (samples currently being analyzed).
  • Ultrasonic wall thickness measurements.
  • Magnetic particle inspection of welds and seams.
  • 3-D laser scanning of the pipe pieces for a digital dimensional record of the evidence.
  • Measurement of the longitudinal and circumferential pup dimensions.
  • Removal of key fracture surfaces from the ruptured segment for further laboratory examination at the NTSB materials lab in Washington.

The following additional work is currently on-going:

  • Precision cleaning of the fracture surfaces on the pieces cut from the ruptured pipe segment.
  • Hardness and microhardness testing.
  • Optical fractographic analysis and photodocumentation of the fracture surfaces on the pieces cut from the ruptured pipe segment.
  • Preliminary scanning electron microscopy of the fracture surfaces on the pieces cut from the ruptured pipe segment

Additional factual updates will be provided and distributed via media advisory as investigative information is developed.


  1. All times mentioned in this report refer to Pacific Daylight Time, unless otherwise specified.

Invitee/Attendee List


  • George Devendorf <>,
  • Brandon Bond <>,
  • Michaela Williams <>,
  • Peter Ohtaki <>,
  • Christiaan Adams <>,
  • Brian Beidelman <>,
  • Clay Sader <>,
  • Jake Fuentes <>,
  • Lisa Costa Sanders <>,
  • Eric Rasmussen <>,
  •, Andrew Brown <>,
  • Dennis Israelski <>,
  • Wendy Schultz <>,
  • Harold Schapelhouman <>,
  • Dennis Haag <>,
  • Rakesh Bharania <>,
  • Catherine Blackadar Nelson <>,
  • “Rob Kelly (rokelly)” <>,
  • Peter Carpenter <>,
  • Harold Brooks <>,
  • Terry Fong <>,
  • Trey Smith <>
  • Mark Lowpensky <>,
  • George Devendorf <>,
  • Jeannie Stamberger <>,
  • Jim Turner <>


At Google

  1. Ryan Zollicofer (Menlo Park Fire Department)
  2. Michaela Williams (American Red Cross)
  3. Harold Brooks (American Red Cross)
  4. George Devendorf (CFI, Fire Marshal and Division Chief, San Bruno Fire)
  5. Jim Wolbrink (San Jose Water, California Utilities “BASIC” and CESA)
  6. Trey Smith (NASA Ames, DMI)
  7. David Coggeshall (Golden Gate Safety Network)
  8. Jim Varner (DMI)
  9. Brian Biedelman (Google)
  10. Christiaan Adams (Google Earth/Crisis Response)
  11. Luke Beckman (InSTEDD) (650) 740-5853
  12. Keith Thode (COO, AidMatrix),
  13. Harold Schapelhouman (Menlo Park Fire)
  14. Christa Taylor (InSTEDD)
  15. Eric Tsay (Menlo Fire)
  16. Ben Chang (Menlo Fire)
  17. Matt Rossi (San Mateo County Sheriff Search and Rescue),
  18. Mark Lowpensky (San Mateo County Sheriff Search and Rescue),, Administrative & Technology Section Chief, Search and Rescue Unit, San Mateo County Sheriff’s Office,
  19. Jake Fuentes (volunteer)
  20. Mike Sena (Northern CA Regional Intelligence Center, www.,
  21. Sumathi Lingappa (self, volunteer mapper)
  22. Peter Ohtaki (Director, California Resiliency Alliance)
  23. Sanjay Waghray (self, volunteer mapper)
  24. Jeannie Stamberger (Carnegie Mellon Silicon Valley Disaster Management Initiative, Associate Director Strategic Programs and Funding)
  25. Andrew Brown (SCEWN/DMI)

On the phone

  1. Rakesh Bahrani (CISCO, NRV van)
  2. Jody Cummings (PG&E)
  3. Elizabeth Proctor (PG&E)
  4. Brandon Oberbauer,, PG&E Sr. GIS Analyst, Pacific Gas & Electric Co., 245 Market St, Mail Code N10A, San Francisco, CA 94105
  5. Katherine Nelson (CISCO, Tactical Operations)
  6. Gregory Smith (Bay Area Chapter, Director for San Bruno incident, American Red Cross)
  7. Eric Park (NASA Ames – GeoCam)
This entry was posted in Disaster Response. Bookmark the permalink.

One Response to San Bruno Fire After Action Report/Technology Debrief

  1. Pingback: EPI and Social Media | Blog | Five Guidelines for Social Media

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s