Monthly Archives: June 2011

Root Cause Analysis - Incident Investigation

Deadly E.Coli Outbreak from Sprouts

By Kim Smiley

Since May, at least 31 people have died and nearly 3,000 have been sickened from E.coli infections in Europe in one of the widest spread and deadliest E.coli outbreaks in recent memory.  After days of confusion, German authorities determined that the source of the contamination is sprouts from an organic farm in northern Germany. The farm has suspended sale of produce and won’t reopen until it is determined safe.

This issue can be investigated by creating a Cause Map, an intuitive format for performing a root cause analysis.  In a Cause Map, the causes contributing to an incident are determined and organized by cause-and-effect relationships.  To view a high level Cause Map of this incident, please click on “Download PDF” above.

This investigation is still underway and additional information can easily be added to the Cause Map as it becomes available. The initial source of contamination at the farm had not yet been determined, but sprouts are known have a high risk of carrying dangerous bacteria.

Sprouts are considered to be a high risk food for a number of reasons.  The seeds are often grown in countries with less stringent inspection criteria so they can arrive at growers already contaminated. Seeds can be contaminated in any number of ways.  E. coli live in the gut of mammals so any time animals or animal waste are near sprout seeds there is a chance of contamination.

It can also be difficult to sanitize the seeds.  Bacteria can hide inside damaged seeds and be missed during sanitizing steps.  Sprouts are also grown in warm water, ideal conditions for growing bacteria as well.  Another factor to consider is that many people eat sprouts raw; cooking would kill any bacteria that were present.

Sprouts have been the source of many bacteria outbreaks in the past.  The U.S. has had at least 30 reported outbreaks related to sprouts in the last 15 years.  Sprouts are associated with enough risk that the Food and Drug Administration has issued warnings for those at high risk, (children, the elderly, pregnant women and people with compromised immune systems) to avoid eating raw sprouts.  If you fall into the high risk category or are just feeling nervous after recent events, the easiest way to prevent bacterial infection from sprouts is to cook them.

Root Cause Analysis - Incident Investigation

Changing the Emergency Response Process

By ThinkReliability Staff

When Line 132 ruptured last September in the community of San Bruno, California, emergency personnel were quick to respond to the natural gas explosion.  The first fire truck was on scene within six minutes of the explosion.  What responders found was a chaotic scene, with multiple wounded and killed and swaths of the neighborhood in flames or simply flattened.  Little did they know that a large natural gas transmission line, feeding the spreading fire, was directly beneath them.  Emergency personnel did their best to clear homes and evacuate the wounded as the fire spread, but the confusion continued for nearly 90 minutes until the gas valves were shut off upstream from the fire.

The subsequent National Transportation Safety Board (NTSB) investigations focused on Pacific Gas and Electric (PG&E) processes following the accident, and found that PG&E was woefully unable to respond quickly to a crisis of this magnitude.  As a set of timelines show, emergency response personnel were already on scene long before PG&E was even aware that a pipeline rupture may be associated with a local fire.  PG&E apparently did not notice an alarm warning them of a pressure drop.  Control systems detected a severe pressure drop approximately four minutes after the disruption; however the PG&E gas control center, located in San Francisco, remained unaware of the explosion and fire until a PG&E dispatch center in Concord called them.  Off duty employees had called-in to the Concord dispatch center 7 and 11 minutes after the incident, alerting them of a large fire in San Bruno.  However it was not until the dispatch center called the gas control center 16 minutes after the explosion that gas control operators realized what was happening.  By this point emergency responders had already arrived at the scene, unaware of the large natural gas pipeline directly under the neighborhood.

What information did emergency responders have as they arrived on scene that day?  Although PG&E itself was aware of the likely service disruption, they failed to notify first responders of any potential danger in those critical minutes after the explosion.  Additionally according to NTSB testimony, the fire department was unaware of the large natural gas pipeline under the community.  Larger transmission pipelines have different operating characteristics than smaller distribution pipelines, including different recommended safety precautions and shut down times.  With a better awareness of the pipeline locations and associated dangers, emergency response personnel could have developed training and response procedures ahead of time for an explosion of this magnitude.  PG&E has since taken steps to enhance its partnership with first responders and other public safety organizations.  Clearly there are other steps that need to be taken as well.

When conducting an investigation, a timeline can be a helpful tool to organize information.  While straightforward to build, timelines can identify areas needing more research and aid in building a process map and a Cause Map.  Compare what happened at PG&E to what emergency responders were doing.  You’ll notice there was a significant delay at PG&E in recognizing there was a problem and then acting upon it.  It took nearly 90 minutes to close valves to shut transmission lines.  Changes must be made to speed up PG&E’s procedures in a crisis situation.

Likewise process maps are a useful tool for determining where a process can use improvement.  In the Current process map, it is noticeable that there are three parallel processes occurring, where information is not being shared in an efficient manner.  The PG&E Dispatch Center only shares information with the Emergency Dispatch Center after they have fully assessed the situation.  This information might come after the fact, as it did in San Bruno, or seriously delay an effective response by EMTs and firefighters.  Going one step further, trained emergency personnel might be able to check with local utilities if they have reason to suspect a natural gas pipeline is involved.  Simple procedural changes, such as who is notified and when, can have significant impacts.

It is important to note that the timeline helps create the most accurate “As Occurred” process map (called Current in this case).  Procedures can differ from actual processes, so it is important to document what actually happened, identify differences in what should have occurred, and figure out why it didn’t.  In this case, PG&E’s procedures were followed and need to be revised.

The NTSB recommendations will undoubtedly lead to multiple changes.  It is easy to focus on material solutions, which tend to be expensive to implement.  Some changes under consideration are the use of remote controlled valves and the replacement of aging pipes.  While there is no doubt that these changes need to happen, other changes can help in the meantime.  Process maps can help identify procedural changes which may be much less expensive, such a modifying notification procedures.

A detailed Cause Map built after the preliminary investigation shows what NTSB investigators believe led the natural gas leak.  More information on the NTSB investigation can be found here.

Root Cause Analysis - Incident Investigation

Great Seattle Fire

By ThinkReliability Staff

On June 6, 1889, a cabinet-maker was heating glue over a gasoline fire.  At about 2:30 p.m., some of the glue boiled over and thus began the greatest fire in Seattle’s history.  We can look at the causes behind this fire in a visual root cause analysis, or Cause Map.  A thorough root cause analysis built as a Cause Map can capture all of the causes in a simple, intuitive format that fits on one page.

First we begin with the impacts to the goals.  There was one confirmed death resulting from the fire, and other fatalities resulting from the cleanup.  These are impacts to the safety goal.  The damage to the surrounding areas can be considered an impact to the environmental goal.  The fire-fighting efforts were insufficient; this can be considered an impact to the customer service goal.  Loss of water and electrical services is an impact to the production goal, the destruction of at least 25 city blocks is an impact to the property goal, and the rebuilding efforts are an impact to the labor goal.

Beginning with these impacted goals, we can lay out the causes of the fire.  The fire did so much damage because of the large area it covered.  It was able to spread over downtown Seattle because it continued to have the three elements required for fire – heat, fuel, and oxygen.  The heat was provided by the initial fire, oxygen by the atmosphere, and plenty of fuel with dry timber buildings.  The weather had been usually dry for the Pacific Northwest, and most of the downtown area had been built with cheap, abundant wood.

Additionally, fire fighters were unable to successfully douse the flames.  The all-volunteer fire department (most of whom reportedly quit after this fire) had insufficient water – hydrants were only placed at every other block, and the water pressure was unable to sustain multiple fire-fighting hoses.  Additionally, some of the water piping was also made of wood, and burned in the fire.  Firefighters attempted to pump water from the nearby bay, but their hoses were not long enough.

Before spreading across the city, the fire spread across the building where it began.  The fire began when glue being heated on a gasoline fire boiled over and lit.  The fire then began to burn the wood chips and turpentine spilled on the floor.  When the worker attempted to spray water at the fire, it only succeeded in spreading the lit turpentine, and thus the fire.  When firefighters arrived, the smoke was so thick that they were unable to find the source of the fire, and so it continued to burn.

The city of Seattle instituted many improvements as a result of this fire.  Wood burnings were banned in the district, and wood pipes were replaced.  A professional fire department was formed, and the city took over the distribution of water.  Possibly because of the vast improvements being made (and maybe because of the reported death of 1 million rats in the fire), the population of Seattle more than doubled in the year after the fire.

View the Cause Map by clicking on “Download PDF” above

Root Cause Analysis - Incident Investigation

Tornado Season of 2011: Worst Ever?

By ThinkReliability Staff

2011 is on pace to be the worst tornado season since record keeping began in 1950.  Communities nationwide have been affected this year, not just those in “Tornado Alley” where twisters are most commonly found.  The marked increase has many wondering just what is going on.  Is it simply greater media attention?  Or perhaps just bad luck this year?  Or maybe this is all because of global warming…

Weather experts agree that it is a combination of factors, but nothing out of the ordinary.  Weather is cyclical, and a higher number of deadly tornados than usual have touched down this year.  Currently 52 deadly tornados have already struck, compared with an annual average of 22.  Additionally these tornados happen to have stuck heavily populated areas.  As recent as April of this year, the EPA has stated that “to date, there is no long-term evidence of systematic changes in [thunderstorms and tornados] over the course of the past 100 years.”

However, some contend that the higher number of tornados must be tied to climate change.  They argue that all the extra energy being stored in the atmosphere is being “expressed in stronger winds…in stronger rainfall.”  How else would it be possible to explain the catastrophic natural phenomenon occurring the last few years?

This is where the Cause Mapping process can help focus all parties on solving the problem, instead of arguing or blaming.  The first step in the process is to define the issue by its impact to overall goals.  In this case, all parties can agree that the destruction and loss of life are the principle impacts.

The next step is to analyze the causes in a visual map.  A simple Cause Map can lay the foundation for a more detailed analysis, so a 5-Why map is usually the best starting point.  From there more causes can be added to the map; all possibilities should be included on the first draft.  When all possible causes are included, it focuses on team on brainstorming instead of debating.

Let’s take a closer look at why so many tornados have hit densely populated areas.  There are primary four reasons identified in the Cause Map.  First, there have been more tornados.  This could be because more are being counted, due to better weather tracking capabilities, or because there simply are more occurring.  Second, there are more forceful tornados than usual.  This could be related to more supercell thunderstorms, since most tornados spring from these types of weather systems.  Because this isn’t known for sure, a question mark indicates that more evidence is needed to support or disprove this hypothesis.  Likewise, it’s possible more strong weather systems are being caused by global warming.

Instead of stopping the analysis to debate global warming, it’s most productive to continue exploring why tornados are touching down in population centers.  It’s not simply a function of the tornados.  There also happen to be more people near where tornados are, and there are more structures which are susceptible to tornado damage.

More people are near where the tornados are because there are more people.  While this is straightforward, it’s often overlooked in the debate and is precisely a reason why more people would perish in a tornado.  People might also be in the area because they have little time to evacuate or take appropriate shelter, unlike in a hurricane.  Advance warning averages just 11 minutes.

Despite many advances in Doppler radar technology and satellite data, tornados are still generally detected the old fashioned way.  Today, a web of 290,000 trained volunteers, called SKYWARN, provide severe weather alert information to the National Weather Service.  Since its inception in the 1970s, SKYWARN has helped the NWS to issue more timely and accurate severe weather warnings.  The NOAA’s National Severe Storms Lab is looking to improve that advanced warning time to 20 minutes, so this might be a possible solution to reducing the number of deaths and injuries caused by tornados.

The fourth factor is that people tend to be located in buildings which are highly susceptible to tornado damage.  More Americans are living in manufactured or modular homes than in previous decades.  As of 2009, there were 8.7 million mobile homes in the United States.  Mobile homes account for nearly half of tornado fatalities.  When other factors are normalized, the data shows unequivocally that mobile homes are more likely to sustain catastrophic damage during tornados.  Some states have begun to take steps to improve the building codes for such dwellings and also to require hardened shelters at mobile home sites.

As even this fairly simple Cause Map demonstrates, there are many factors contributing to this season’s frightening weather.  Focusing on a single cause can mask the many reasons required to produce an effect, and in the end only limits productive debate.