Root Cause Analysis Tools
Thread Starter
Join Date: Oct 2005
Location: UK/Philippines/Italy
Age: 73
Posts: 557
Likes: 0
Received 0 Likes
on
0 Posts
Root Cause Analysis Tools
Does anyone have any direct experience of using Tap Root as a RCA tool? If so, I would greatly appreciate a view on its efficacy.
Join Date: Jan 2010
Location: Germany
Age: 43
Posts: 25
Likes: 0
Received 0 Likes
on
0 Posts
Just one comment: I guess, that root cause analysis is no more considered as state-of-the-art?
Last weekend I worked throug the Colgan Air Accident report (NTSB) and there I really experienced that there was no root cause. There were so many important fragements, but no single root cause.
But sorry, this was off-topic!
Last weekend I worked throug the Colgan Air Accident report (NTSB) and there I really experienced that there was no root cause. There were so many important fragements, but no single root cause.
But sorry, this was off-topic!
Join Date: May 2008
Location: Kettering
Posts: 150
Likes: 0
Received 0 Likes
on
0 Posts
Apollo30
Root Cause Analysis is just a generic name for a cluster of tools. It is used extensively, if badly, in the UK Health Service. The individual components are still used in aviation investigation - even if they are not called 'RCA'.
The problem with 'root cause' is deciding when to stop asking questions. And in very messy accidents like Colgan there will be multiple 'root causes' each underpinning a particular influencing factor.
For example, the use of a third party training provider did not cause the crash. But such companies tend to train to the minimum necessary to meet the check ride. Companies can actually ask for more than the minimum and they can also ask for the 'one size fits all' course to be tailored to company needs ... but this costs money. So, here is a thread that could be explored to establish some 'root causes' of the influencing factor but which would not be considered a root cause of the accident.
Root Cause Analysis is just a generic name for a cluster of tools. It is used extensively, if badly, in the UK Health Service. The individual components are still used in aviation investigation - even if they are not called 'RCA'.
The problem with 'root cause' is deciding when to stop asking questions. And in very messy accidents like Colgan there will be multiple 'root causes' each underpinning a particular influencing factor.
For example, the use of a third party training provider did not cause the crash. But such companies tend to train to the minimum necessary to meet the check ride. Companies can actually ask for more than the minimum and they can also ask for the 'one size fits all' course to be tailored to company needs ... but this costs money. So, here is a thread that could be explored to establish some 'root causes' of the influencing factor but which would not be considered a root cause of the accident.
Join Date: May 2008
Location: Australia
Posts: 89
Likes: 0
Received 0 Likes
on
0 Posts
When I took a look some time ago at accident investigation reports, and noted the average report identified seven causal factors.
It is very rare than one could point at a single factor and say that caused the accident.
It highlights the systemic nature of accidents and I used it in my report to point out the fallacy of "blame the pilot" culture.
hugel
It is very rare than one could point at a single factor and say that caused the accident.
It highlights the systemic nature of accidents and I used it in my report to point out the fallacy of "blame the pilot" culture.
hugel
Join Date: May 2008
Location: Kettering
Posts: 150
Likes: 0
Received 0 Likes
on
0 Posts
Hugel,
I would have thought that the 'blame the pilot' fallacy is well-recognised in most 'enlightened' territories even if not always fully endorsed through the actions of society, management and investigators.
I think the problem with Root Cause Analysis is the title. As I said in my previous post, it isn't a single technique and it doesn't necessarily get to 'root cause'. You need to use the appropriate selection of techniques to get past 'proximal' cause in order to get an accurate understanding of why things happened.
We can, in fact, get to a single factor that causes an accident. Usually that single factor is the action sequence of the crew immediately prior to impact. A different action sequence would have resulted in a different outcome. However, first, we need to avoid 'blame' and second (most importantly) we need to understand why the crew through that their actions were appropriate under the circumstances. This is what the 'root cause repertoire' is supposed to clarify.
I would have thought that the 'blame the pilot' fallacy is well-recognised in most 'enlightened' territories even if not always fully endorsed through the actions of society, management and investigators.
I think the problem with Root Cause Analysis is the title. As I said in my previous post, it isn't a single technique and it doesn't necessarily get to 'root cause'. You need to use the appropriate selection of techniques to get past 'proximal' cause in order to get an accurate understanding of why things happened.
We can, in fact, get to a single factor that causes an accident. Usually that single factor is the action sequence of the crew immediately prior to impact. A different action sequence would have resulted in a different outcome. However, first, we need to avoid 'blame' and second (most importantly) we need to understand why the crew through that their actions were appropriate under the circumstances. This is what the 'root cause repertoire' is supposed to clarify.
There are no limits
Join Date: May 2002
Location: Shrewsbury, England.
Age: 67
Posts: 505
Received 0 Likes
on
0 Posts
Here is anexample of how TapRooT can be used, from the Flight Safety Foundation
http://flightsafety.org/files/TapRooT_application.pdf
To me the crux of any RCA has to be understanding the Problem Statement or Definition to start with.
I haven't studied Colgan in depth, but my understanding of the cause was that the pilot made an incorrect control input in response to the incipient stall warning. That was the cause and all the other factors were 'contributing'.
http://flightsafety.org/files/TapRooT_application.pdf
To me the crux of any RCA has to be understanding the Problem Statement or Definition to start with.
I haven't studied Colgan in depth, but my understanding of the cause was that the pilot made an incorrect control input in response to the incipient stall warning. That was the cause and all the other factors were 'contributing'.
Thread Starter
Join Date: Oct 2005
Location: UK/Philippines/Italy
Age: 73
Posts: 557
Likes: 0
Received 0 Likes
on
0 Posts
Many thanks to all for your responses. A particular thanks to What Limits for providing the specific example. My view is that Tap Root is quite useful especially for trouble shooting equipment failures.
My background is that of an ex military pilot who now works with the petrochemical industry. I still fly privately but have no commercial experience. A good proportion of my work involves RCA analysis of 'events' and advising on how to fill in the holes in the Swiss cheese model. I study aircraft accidents as they are some of the most investigated accidents there are.
In essence, all RCA techniques are based on the 5 'Whys'. Why Tree, Why Root, Drill Down etc all use the same basic techniques. Some require a skilled facilitator (me
) to keep things focused.
My view is that you can now go down to 7 'whys' if necessary.
I also define 'root cause': 'The lowest level system failure that - if fixed - would have prevented the event'.
I agree that there can - sometimes - be more than one of these. There is, more often, multiple causal factors, contributing factors; call them what you wish. However, if you were to accept my definition, you would probably arrive at only one root cause and multiple CFs.
I like to split the analysis into three stages:
1. The 'What Tree': this analyses what happened.
2. The 'Who Tree': this gets to who dis what and when. Often difficult because of defensive attitudes and turf wars.
3.Keep asking 'why' and, with skill, you get to the 'why' at a systems level. This is the bit you fix.
This is at the 'who' level. If you were to ask yourself why the Captain made inappropriate control inputs you will identify a number of CFs. You need to get past the actual actions to the systems level failure. Start from the POV that most people try to do the right thing.
Apollo 30:
I have read the report. What? 40 plus recommendations. The Probable Cause (not root cause) was the inappropriate reaction to the stick shaker.There are 4 CFs listed.
What is also clear in the report is the NTSB's total frustration with the failure of the FAA to implement earlier recommendations. Indeed, if you were to look a deeper root causes, this might be one!
However, based on my earlier definition, you could have the root cause of this accident being Colgan's failure to ensure that pilots had the requisite skills to do the job. This might look at - in this case - the hiring process.
Totally agree. Get that wrong and you don't stand a chance!!!! You have to analyse the right thing or try to solve the right problem! Many examples of that including the NASA pen (fairy) tale.
Thanks. Interesting discussion.
My background is that of an ex military pilot who now works with the petrochemical industry. I still fly privately but have no commercial experience. A good proportion of my work involves RCA analysis of 'events' and advising on how to fill in the holes in the Swiss cheese model. I study aircraft accidents as they are some of the most investigated accidents there are.
In essence, all RCA techniques are based on the 5 'Whys'. Why Tree, Why Root, Drill Down etc all use the same basic techniques. Some require a skilled facilitator (me
) to keep things focused.My view is that you can now go down to 7 'whys' if necessary.
I also define 'root cause': 'The lowest level system failure that - if fixed - would have prevented the event'.
I agree that there can - sometimes - be more than one of these. There is, more often, multiple causal factors, contributing factors; call them what you wish. However, if you were to accept my definition, you would probably arrive at only one root cause and multiple CFs.
I like to split the analysis into three stages:
1. The 'What Tree': this analyses what happened.
2. The 'Who Tree': this gets to who dis what and when. Often difficult because of defensive attitudes and turf wars.
3.Keep asking 'why' and, with skill, you get to the 'why' at a systems level. This is the bit you fix.
Usually that single factor is the action sequence of the crew immediately prior to impact.
Apollo 30:
Last weekend I worked throug the Colgan Air Accident report (NTSB) and there I really experienced that there was no root cause. There were so many important fragments, but no single root cause.
What is also clear in the report is the NTSB's total frustration with the failure of the FAA to implement earlier recommendations. Indeed, if you were to look a deeper root causes, this might be one!
However, based on my earlier definition, you could have the root cause of this accident being Colgan's failure to ensure that pilots had the requisite skills to do the job. This might look at - in this case - the hiring process.
To me the crux of any RCA has to be understanding the Problem Statement or Definition to start with.
Thanks. Interesting discussion.
Join Date: May 2008
Location: Kettering
Posts: 150
Likes: 0
Received 0 Likes
on
0 Posts
"Quote:
Usually that single factor is the action sequence of the crew immediately prior to impact.
This is at the 'who' level. If you were to ask yourself why the Captain made inappropriate control inputs you will identify a number of CFs. You need to get past the actual actions to the systems level failure"
I think that's what I said. I was trying to distinguish between proximal cause and distal, or root, cause. The problem we have now is that 'no single cause' has become a mantra and actually distracts us from looking more closely at what the actors actually did that contributed to failure.
However, we also need to remember that it's not what they did, it's why they did it that is the most important point to understand.
Usually that single factor is the action sequence of the crew immediately prior to impact.
This is at the 'who' level. If you were to ask yourself why the Captain made inappropriate control inputs you will identify a number of CFs. You need to get past the actual actions to the systems level failure"
I think that's what I said. I was trying to distinguish between proximal cause and distal, or root, cause. The problem we have now is that 'no single cause' has become a mantra and actually distracts us from looking more closely at what the actors actually did that contributed to failure.
However, we also need to remember that it's not what they did, it's why they did it that is the most important point to understand.
Thread Starter
Join Date: Oct 2005
Location: UK/Philippines/Italy
Age: 73
Posts: 557
Likes: 0
Received 0 Likes
on
0 Posts
Sorry, no intention to criticise your post which was well thought out and expressed. I was trying to use part of it to illustrate what I call the 'who' level at which these things tend to stop. 'The pilot failed to lower the wheels' level.
We agree totally.
Also, go back to my definition. The word 'system' is important.
On a broader note, you could find root cause in the NTSB's lack of ability to be able to enforce its recommendations on the FAA. Not surprising, as I understand it, the NTSB does not have these powers. Dig deeper and you may have an interdepartmental government failure!
And for goodness sake please don't start me with the military court martial, board of inquiry system. Both guaranteed to get to the root cause.
it's not what they did, it's why they did it that is the most important point to understand.
Also, go back to my definition. The word 'system' is important.
On a broader note, you could find root cause in the NTSB's lack of ability to be able to enforce its recommendations on the FAA. Not surprising, as I understand it, the NTSB does not have these powers. Dig deeper and you may have an interdepartmental government failure!
And for goodness sake please don't start me with the military court martial, board of inquiry system. Both guaranteed to get to the root cause.
Join Date: Jan 2010
Location: Germany
Age: 43
Posts: 25
Likes: 0
Received 0 Likes
on
0 Posts
When working in the field of accicent analysis - what NTSB does - you are always confronted with legal aspects, liability, recommendations and so on.
Another very interesting approach is done by Hollnagel with the FRAM (functionl resonance analysis method):
fram (erikhollnagel2) or http://www.skybrary.aero/bookshelf/books/402.pdf
In my point of view this method gives a very deep view into sociotechnical systems even it's not too simple to apply. And I think it also fits a litte bit more to the idea that often there are many "root causes". For me, I define a "root cause" as one single thing, so for me this concept is not sufficient to describe several contributors which lead in common to an accident.
Another very interesting approach is done by Hollnagel with the FRAM (functionl resonance analysis method):
fram (erikhollnagel2) or http://www.skybrary.aero/bookshelf/books/402.pdf
In my point of view this method gives a very deep view into sociotechnical systems even it's not too simple to apply. And I think it also fits a litte bit more to the idea that often there are many "root causes". For me, I define a "root cause" as one single thing, so for me this concept is not sufficient to describe several contributors which lead in common to an accident.
