Human Error Mistakes are the greatest remaining threat to your plant and equipment reliability
Once you understand how human error mistakes occur you can design processes to prevent their occurrence—you can error-proof work and eliminate the consequences of people making mistakes
Human error mistakes will always happen. Humans are not robots—we are flesh and bone animals limited by the capabilities of our body design and our bodily and emotional needs.
PEW/PWW EAM Course Day 1 – Foundations Session 6 – Human Error
Common Wrongs Humans Do To Machines
We twist them
We squeeze them
We hit them
We poison them
We burn them
We shake them
We break them
We choke them
We overload them
We boil them
The Odds are Against Doing it Right!
There are plenty of opportunities to make human error mistakes in aviation maintenance. Look at this easy example of one bolt with many nuts. This nut and bolt assembly can only be disassembled one way. However, there are over 40,000 combinations for reassembly, of which only one is correct! This is a good example of the complexity and the chance of making human error mistakes in every maintenance environment.
3 Extracts on the Causes of Equipment Failure
The extracts are from three sources investigating industrial plant and equipment failures. They all confirm that human error mistakes are involved with the causes of 80% – 90% of operating equipment failures.
The aircraft and their systems are improving. Reducing human error mistakes is one of the very last “safety frontiers.”
Human Factors – limitations of Flesh and Bone
This slide lists the human factors, where they occur and how they cause humans to make errors. We all make the same kind of errors and we are all human. There are level 3 details for: fatigue, anthropometrics (the study of human dimensions), and sensing and perception examples.
The 12 Most Common Causes of Human Error
You can just talk about the dirty dozen or play the animation. Talk may be better.
“To err is human.” Throughout the human factors training you may hear the words “Dirty Dozen.“ The dirty dozen is a listing of the 12 most common causes of human error in maintenance. The concept was developed by Mr. Gordon Dupont at Transport Canada. If we could eliminate or control these 12 causes of human error mistakes we would eliminate a very high percentage of maintenance-related events. We will look at each factor that contributes to an error, and offer possible corrective actions to prevent such error. The following is a listing of the Dirty Dozen:
Lack of Communication;
Lack of Teamwork;
Norms, i.e. culture;
Lack of Knowledge;
Lack of Awareness;
Lack of Resources;
Lack of Assertiveness;
By the end of your human factors training, you should remember many of these errors, and also know the best methods for preventing them.
PEW SOLUTION: Apply the Answers in the Human Error Rate Table to Reduce Human Error Mistakes 10,000%
The Human Error Rate Table confirms that ‘human element’ error is real and unavoidable. We do not perform well when tasks are structured in ways that require care, and we perform especially badly under complicated non-routine conditions. Add stress into that that mix and you get disaster. It’s very easy to create situations that promote human error mistakes.
Machines Suffer High Risk from Human Contact
From our knowledge of series reliability we know that for a machine to be highly reliable each part in the machine must be even more highly reliable. Yet we maintain our machines using work processes of poor reliability where human error mistakes can easily happen. It is just like the DuPont defect creation model—our tasks have great chance for causing defects that will eventually cause us to do more work and cost money. When your people work on machines you need to ensure that the work is done using task activities that are highly reliable, and the people doing the task know and understand why they need to double-check every thing they do.
Reliability of Series Work Process
When work is done in series, the poor performance of any task leads to a poor outcome for the whole procedure. In a long series process there is great opportunity for error and defect introduction.
In a 12 task maintenance procedure, if all but two tasks are always done 100% right and the outstanding two are done right only 90% of the time, then the certainty that the whole procedure is right whenever it is done drops to 81 times out of 100 (81%). If instead all 12 tasks are likely to be done right 90% of the time (10% of the time they will be done wrong), the likelihood that the whole procedure will be done right is only 28% of the time. Should the chance of each 12 tasks being done right rise to 99%, then the chance of the whole procedure being done right at any time becomes 88.6%.
The only way to ensure that series work always is done 100% right every time it is performed is to ensure every task is done 100% right every time.
Carpenter’s creed: ‘measure twice, cut once’
An example of a parallel process is the carpenter’s creed, ‘Measure twice; cut once’. Carpenters know that the double-check will save problems and trouble later. We can turn the adage into the simple parallel process shown in the bottom figure.
The typical error rate in reading a tape measure is five times in every thousand it will be misread, or 995 times out of 1000 it will be right (a reliability of 0.995). The carpenter will cut the wood in the wrong spot about once every 200 times. It is not hard to imagine a carpenter doing 50 cuts a day. So about once a working week they would cut the wood in the wrong place and have to throw it away. When he also does the proof-test measure the chance of getting the cut right rises to 0.9998, which is an error rate of 2 in every 10,000 times. With 50 cuts a day they will make an error once every 100 working days, or about every 20 working weeks. The simple addition of a check-test produced twenty times fewer measurement mistakes. That is the power of paralleling test activities to tasks to ensure they are right.
PEW SOLUTION: Use the power of parallel proof-tests on every task activity
The Figure shows the 5-task series process as a paralleled 5-task process. Each task includes a parallel proof-test activity to confirm the task is correct. Exactly like the carpenter’s creed, ‘measure twice, cut once’. If we take the 0.9 reliability of maintenance work for each task, and for the inspect-and-measure proof-test increase it to 0.99 (because testing is carefully done using high quality tools and procedures), then the reliability of each parallel-tested step is:
Rtask = 1 – [(1- R1) x (1- R1t)]
= 1 – [(1-0.9) x (1-0.99)] = 1 – [(0.1) x (0.01)] = 1 – [0.001]
= 0.999 (99.9%)
By combining a task with a test activity to prove that the task is right, we create a highly reliable task. Add proof-test activities to all tasks in our 5-step job and you create a high-reliability work process. The reliability of the entire job is now:
Rjob = 0.999 x 0.999 x 0.999 x 0.999 x 0.999 = 0.995 (i.e. 99.5%)
Paralleling a proof-test to each task drives the reliability for the entire job to its highest level yet. But even 0.995 reliability means that 5 times out of every 1000 opportunities, the job will be wrong. In a large, busy operation with many people, one thousand opportunities for error accrue rapidly. Similarly, where numerous processes are used to make a product there is hundreds, even thousands, of opportunities a day for error to happen along the process chain. We need job and process reliability of great certainty if we want excellence in our businesses. You can achieve this by continuing the paralleling activity with each task.
Quite literally, you can design the reliability that you want into a job or into a machine. To have high-reliability work processes, build parallel inspection activities into the performance of the work. The activity of doing the work must ensure that high-reliability is the natural outcome. People must know for themselves when a thing is ‘done right’. Make proof-testing a standard practice in the system of work; make it ‘the way we do things around here’. Parallel all critical tasks done in a job with very specific and certain error-preventing tests. Then you can be sure that the process is able to deliver the quality you want.
Can we get 10,000% fewer errors? | Parallel Proof Test Activity Reliability
Most organisations without a properly working quality management system have error rates of 15 or more in 100. That means for every 100 opportunities for things to go wrong at least 15 do. They are two-and-a-half (2-1/2) sigma operations. Many of these would have error rates of 30 in 100 and more; a two sigma operation. Such a situation is fantastically expensive for a company because a lot of work will have to be done twice. Very few companies can afford to pay twice as much for a job as they expected to pay.
The Maintenance Planner is in a situation of being able to lift an organisation’s performance because they can control the quality of information and the work standards being given to the maintenance crew so that the number of human error mistakes are reduced.
The Maintenance Planner can introduce a proof test into each task in a procedure to ensure the task is done right. If all tasks are done right the outcome is right.
Any time we can turn a series task into a parallel task we increase the chance of it being done right.
In the slide, each task is turned into a parallel task by adding a 3T test, which is itself also done right only 90% of the time. Where we would have got the procedure right only 28% of the time if it was a series process, we get it right 88.6% of the time when we have a parallel test, even though both the task and the test are each done right only 90% of the time.
Clearly adding a test check into each task makes tremendous improvement in workmanship quality.
Carpenter’s creed: ‘measure twice, cut once’ — YOU MUST DO INDEPENDENT MEASUREMENTS
Beware that if you want high levels of control, each layer of proof test needs to be independent of the others so that any errors in one test is not duplicated in another. For example, if you measure a second time, and the second test is done with the same measuring device used for the first test ,you will commit a ‘common error ‘. The measuring device may be wrong and the second test would not detect that error. Another example is if a second inspection is used as a proof-test,but it is done by another person with the same mistaken understanding as the first person.
But Parallel Test Tasks adds Cost… So what can you do?
Once you know that maintenance is all about providing reliability, AND you understand that reliability is all about removing the chance of things going wrong, you are able to make necessary changes and improvements to your processes.
If you want a higher likelihood that the outcome will be right, you must provide additional risk control at every step in the process used to deliver that outcome. Then, as the chance of human error mistakes in individual tasks is reduced, the likelihood that the whole job will be done right rises.
The new Industrial and Manufacturing Wellness book contains all the latest information, all the latest templates, and worked examples of how to design and build a Plant Wellness Way Enterprise Asset Management (PWWEAM) system-of-reliability. Get the book from its publisher, Industrial Press, and Amazon Books.
The PLANT WELLNESS WAY EAM TRAINING COURSE teaches you to use and master the Plant Wellness Way EAM methodology. Follow this link to read about Training for New Users in the Plant Wellness Way EAM Methodology for World Class Reliability.
You are welcome to go to the Plant Wellness Way Tutorials webpage and look at worked examples of Plant Wellness Way EAM techniques and read in-depth explanations of the latest version of many PWWEAM presentation slides.
Use the head office email address on the Contact Us page if you have questions about the videos.