As the Boeing 787 Dreamliner fleet is grounded
(February 25-th, 2013), it is curious to read the numerous articles
reporting on the lithium-ion battery problem that is presumably
responsible for the grounding. Clearly, the investigation is still under
way and it would be foolish to make any statements as to what has
happened and how it will affect the 787 and Boeing. That is not to goal
of this blog. The objective is to draw attention to another issue,
namely that of design and operation of super-complex and sophisticated
systems, such as the 787 (or the A330, or the economy).
High complexity regimes is uncharted territory. While most people struggle with definitions of complexity (many still think it has to do with ants or storms of starlings) it is hitting us hard on many fronts. In the past few decades we have conceived, designed and constructed extremely complex systems and infrastructures on which our lives depend to a very large degree. The list is endless but it all comes down to huge networks, systems of systems, computers, software, information and communication, the internet, etc. Such systems are indeed very complex and yet they have been designed without taking complexity into account. This is paradoxical to say the least. Complexity is the hallmark of our times and yet complexity is not a design-attribute when it comes to putting in place sophisticated products, systems or infrastructures. Engineers should be the first to be concerned with measuring complexity and keeping it under control in any manufacturing project, just like it happens for costs and schedules. What happens if you throw complexity under the carpet is that it will strike back in the form of guaranteed cost overruns and late delivery. The A380 and the 787 Dreamliner immediately come to mind. And then, once the product is on the market, high complexity will strike again in the form of fragility. Just think of how sophisticated financial products have run out of control and what they have done to our global economy.
Designing highly complex and sophisticated products while taking complexity into account has been described in our old blog. Today, there is a cloud-based tool for doing precisely that. Visit the Design4Resilience website.
However, it is even more shocking to discover how linear thinking still dominates our thinking and philosophy. Unfortunately this applies to many engineers too. In a recent article on the 787 battery problem we read:
".... One key question for safety investigators is how the battery's eight individual cells became volatile even though the overall voltage to the battery was steady and didn't exceed the 32-volt capacity, officials have said."
The logic is the following:
There could be nothing more wrong. The above reasoning applies to linear systems only (and even then you still need to be careful). In Nature very few things are genuinely linear (the big question is why universities teach how to fiercely linearize everything). In reality you don't need to take things to the extreme to get yourself into trouble. You can run into very unfavorable circumstances even when system parameters (state vector components) combine and interact far from extreme design values. At school they teach you that if epsilon is small then epsilon-squared is even smaller and may be neglected. Yes, that's how Taylor series work but imagine that there are thousands of epsilons within your system. What if they suddenly combine and push in the same direction. How can you dismiss that? You cannot. And if you do the consequences will be more severe if your system is highly complex. Moreover, finding the problem will be more demanding. Especially if your system is run by computers and millions of lines of real-time code. Sounds familiar?
Even with a very coarse description, whereby a few tens of variables describe a single subsystem, the number of possible interactions is in the range of thousands. It is impossible to analyze them all. What couples apparently uncoupled systems is omnipresent software.
It is possible to build sophisticated products and systems but there are ways to make them less fragile. Make them simpler. Simplicity is the ultimate sophistication (Leonardo da Vinci).
Click here and move the mouse.
Update (January 28-th, 2013). Latest reports point to a potentially systemic issue:
"I think people had their fingers crossed that it was a battery fault... it looks more systemic and serious to me. I suspect it could be difficult to identify the cause,"
Update (February 4-th). Article speaking of how supply chain complexity was a potential source of Boeing's trouble with the Dreamliner.
www.ontonix.com
High complexity regimes is uncharted territory. While most people struggle with definitions of complexity (many still think it has to do with ants or storms of starlings) it is hitting us hard on many fronts. In the past few decades we have conceived, designed and constructed extremely complex systems and infrastructures on which our lives depend to a very large degree. The list is endless but it all comes down to huge networks, systems of systems, computers, software, information and communication, the internet, etc. Such systems are indeed very complex and yet they have been designed without taking complexity into account. This is paradoxical to say the least. Complexity is the hallmark of our times and yet complexity is not a design-attribute when it comes to putting in place sophisticated products, systems or infrastructures. Engineers should be the first to be concerned with measuring complexity and keeping it under control in any manufacturing project, just like it happens for costs and schedules. What happens if you throw complexity under the carpet is that it will strike back in the form of guaranteed cost overruns and late delivery. The A380 and the 787 Dreamliner immediately come to mind. And then, once the product is on the market, high complexity will strike again in the form of fragility. Just think of how sophisticated financial products have run out of control and what they have done to our global economy.
Designing highly complex and sophisticated products while taking complexity into account has been described in our old blog. Today, there is a cloud-based tool for doing precisely that. Visit the Design4Resilience website.
However, it is even more shocking to discover how linear thinking still dominates our thinking and philosophy. Unfortunately this applies to many engineers too. In a recent article on the 787 battery problem we read:
".... One key question for safety investigators is how the battery's eight individual cells became volatile even though the overall voltage to the battery was steady and didn't exceed the 32-volt capacity, officials have said."
The logic is the following:
- A system is monitored by checking a certain number of parameters (hundreds, thousands).
- If each parmeter is within bounds then all is OK.
- The worst case is when all (or many) parameters reach the end of the scale.
There could be nothing more wrong. The above reasoning applies to linear systems only (and even then you still need to be careful). In Nature very few things are genuinely linear (the big question is why universities teach how to fiercely linearize everything). In reality you don't need to take things to the extreme to get yourself into trouble. You can run into very unfavorable circumstances even when system parameters (state vector components) combine and interact far from extreme design values. At school they teach you that if epsilon is small then epsilon-squared is even smaller and may be neglected. Yes, that's how Taylor series work but imagine that there are thousands of epsilons within your system. What if they suddenly combine and push in the same direction. How can you dismiss that? You cannot. And if you do the consequences will be more severe if your system is highly complex. Moreover, finding the problem will be more demanding. Especially if your system is run by computers and millions of lines of real-time code. Sounds familiar?
The laws
of systemantics are said to be pseudo-science. A few of these laws
are listed below. They apply to highly complex systems. Think of these
laws and then think of the super-complex systems and infrastructures on
which our lives depend.
- Le Chatelier’s Principle: Complex systems tend to oppose their own proper function. As systems grow in complexity, they tend to oppose their stated function. (This law makes you think of the Eurozone).
- A complex system cannot be “made” to work. It either works or it doesn’t.
- A complex system that works is invariably found to have evolved from a simple system that works.
- A complex system designed from scratch never works and cannot be patched up to make it work. You have to start over, beginning with a working simple system.
- The Functional Indeterminacy Theorem (F.I.T.): In complex systems, malfunction and even total non-function may not be detectable for long periods, if ever.
- The Fundamental Failure-Mode Theorem (F.F.T.): Complex systems usually operate in failure mode.
- A complex system can fail in an infinite number of ways.
- The larger the system, the greater the probability of unexpected failure.
- As systems grow in size, they tend to lose basic functions.
- Colossal systems foster colossal errors.
Even with a very coarse description, whereby a few tens of variables describe a single subsystem, the number of possible interactions is in the range of thousands. It is impossible to analyze them all. What couples apparently uncoupled systems is omnipresent software.
It is possible to build sophisticated products and systems but there are ways to make them less fragile. Make them simpler. Simplicity is the ultimate sophistication (Leonardo da Vinci).
Click here and move the mouse.
Update (January 28-th, 2013). Latest reports point to a potentially systemic issue:
"I think people had their fingers crossed that it was a battery fault... it looks more systemic and serious to me. I suspect it could be difficult to identify the cause,"
Update (February 4-th). Article speaking of how supply chain complexity was a potential source of Boeing's trouble with the Dreamliner.
www.ontonix.com
No comments:
Post a Comment