To celebrate National Engineers Week (February 19-25), MIT Press has conducted a series of Q&As with the authors of the new Engineering Systems series books.
The third and final Q&A for National Engineers Week 2012 is with Richard de Neufville, author (with Stefan Scholtes) of Flexibility in Engineering Design. He is Professor of Engineering Systems and Civil and Environmental Engineering at MIT. He was Founding Chairman of the MIT Technology and Policy Program.
Q. What is engineering design?
A. To my way of thinking, engineering design is the process of creating technically sound products that provide good value for money.
Good engineering design will get the technology right — the thing will work. But that is not enough! As an early engineering educator colorfully put it, "an engineer is a person who can do for a buck what any damn fool can do for two." Good engineering design delivers good value.
Flexibility helps deliver good value by dealing with the reality that as we cannot predict the future, we must adapt to it.
Q. How does flexibility and engineering design relate to engineering systems?
A. Today, we mostly produce engineering systems — complex combinations of different kinds of engineered products.
These may be big—electrical grids serving many customers from several power plants. They may also be small—computer chips with hundreds of elements.
The modern challenge for engineers is how we design these complex assemblages of parts and capabilities. We need to deal with the network of interactions at a scale that we are not accustomed to. This is what engineering systems design is about.
Flexibility is fundamental to good engineering systems design because of the inescapable uncertainty surrounding the use and performance of engineering systems. We cannot predict exactly how these systems will function, and we cannot know how they will be used. So we need to be able to adapt to actual circumstances.
Q. What is "forecast uncertainty"?
A. Simply put, we are not omniscient and cannot predict the future. We cannot be sure of what will happen. Our forecasts are thus inescapably uncertain. We need to deal with this fundamental reality.
Flexibility, the ability to adapt to the range of future conditions, is by definition an important way to deal with forecast uncertainty. If you can't fix the future, learn to adapt and live with it.
Q. Why are forecasts fundamental to engineering design?
A. Design, engineering design in particular, is about creating products for future use and enjoyment.
Picking up on the idea that good design provides good value for the money or effort expended, we want to shape and size our designs to what people will value. Good design thus implicitly relies on forecasts of what might eventually be needed or desired.
For example, good design will not create a "bridge to nowhere". The bridge itself might be structurally and technically sound. But if it serves no purpose, it is a waste of money and not good design.
Q. In Chapter 3, you state that "flexibility in design routinely improves performance by 25 percent or more." Can you give an example of how design flexibility can increase a project's value?
A. Design flexibility improves performance by allowing us to adapt products to future circumstances and thus to obtain good performance under a range of situations. A product without design flexibility, without the capacity to adapt, often leaves us with the wrong thing at the wrong time.
The book makes the point with the simple example of the design for a multi-story parking garage.
A standard design delivers a facility with a fixed number of levels and is likely to be too big at the beginning, and then too small when traffic builds up. However, a flexible design starts small and expands if and when needed. The flexible design thus avoids losses, and takes advantages of upside opportunities. It thus can provide a "win-win" product. Case after case show that the expected benefits from flexible design are truly significant.
Q. What are the different kinds of flexibility for a system and how do engineers determine which flexibilities will add the most value?
A. An obvious kind of flexibility concerns the size of a product – can we easily expand it if needed?
For example, can we double-deck a bridge to carry more traffic, as we did for the George Washington Bridge over the Hudson River?
Another flexibility concerns capability— can we upgrade it? An example is replacing a component with a new, better version, as aircraft designers routinely do when they modernize aircraft with more powerful engines or better computers.
A different flexibility enables changes in function — can we re-purpose a product to satisfy different needs, as we do when we transform office buildings into apartment houses?
Identifying which flexibility might add the most value is a challenge! The book discusses this process. As the value of flexibility lies in the capacity to adapt to the range of futures, we first need to focus on what the range of futures might be. We then need to analyze the possible flexibilities to identify those that add the most value. The choice of flexibility thus depends on the dynamics of the context. What is right for a high-tech product may not be best for conventional construction. What is appropriate for Singapore might not be best for the United States.
Q. Though it's impossible to predict the future, what are the key factors that engineers should consider when designing a project or system?
A. The key idea is that "the forecast is 'always wrong'"! Designers need to recognize the great likelihood of unforeseen developments that affect the performance of their products. Economic cycles alter demand, new technologies create new opportunities, and political events drive industries and markets. We need to be modest about our ability to specify requirements for a design. We thus need to recognize the great need for flexibility to adapt to future realities.
Q. Can you describe the common obstacles that engineers face when trying to implement design flexibility?
A. A most common obstacle is the refrain that "flexibility is nice, but we can't afford it".
This mantra comes from folks who think they know exactly what is needed, and who fail to recognize the ranges of conditions for their designs. It reflects a mental block that turns a blind eye to the reality of uncertainty about needs.
Q. What is the "Flaw of Averages"?
A. The "flaw of averages" is the mistake (the flaw) of believing that everything will average out (the "law" of averages). Designers make this mistake when they design products around "average" or "most likely" conditions — rather than design for the range of possible futures. When designers design around averages, they systematically estimate the value of their designs incorrectly, and make the wrong design choices.
The "flaw of averages" is a very simple and long-established mathematical fact. (It's also called "Jensen's law" and you can Google it for details.) Unfortunately, standard design processes fixed on a single set of requirements routinely fall into this error.
Q. What kinds of changes (if any) do you think we need to make in engineering education?
A. As regards design, the most important change is for educators to open our students' minds to the reality that forecasts of what is needed are imprecise at best. When we recognize this reality, we re-frame the design problem from meeting some fixed set of requirements to that of being able to adapt to meeting a range of requirements. The rest follows!
The third and final Q&A for National Engineers Week 2012 is with Richard de Neufville, author (with Stefan Scholtes) of Flexibility in Engineering Design. He is Professor of Engineering Systems and Civil and Environmental Engineering at MIT. He was Founding Chairman of the MIT Technology and Policy Program.
Q. What is engineering design?
A. To my way of thinking, engineering design is the process of creating technically sound products that provide good value for money.
Good engineering design will get the technology right — the thing will work. But that is not enough! As an early engineering educator colorfully put it, "an engineer is a person who can do for a buck what any damn fool can do for two." Good engineering design delivers good value.
Flexibility helps deliver good value by dealing with the reality that as we cannot predict the future, we must adapt to it.
Q. How does flexibility and engineering design relate to engineering systems?
A. Today, we mostly produce engineering systems — complex combinations of different kinds of engineered products.
These may be big—electrical grids serving many customers from several power plants. They may also be small—computer chips with hundreds of elements.
The modern challenge for engineers is how we design these complex assemblages of parts and capabilities. We need to deal with the network of interactions at a scale that we are not accustomed to. This is what engineering systems design is about.
Flexibility is fundamental to good engineering systems design because of the inescapable uncertainty surrounding the use and performance of engineering systems. We cannot predict exactly how these systems will function, and we cannot know how they will be used. So we need to be able to adapt to actual circumstances.
Q. What is "forecast uncertainty"?
A. Simply put, we are not omniscient and cannot predict the future. We cannot be sure of what will happen. Our forecasts are thus inescapably uncertain. We need to deal with this fundamental reality.
Flexibility, the ability to adapt to the range of future conditions, is by definition an important way to deal with forecast uncertainty. If you can't fix the future, learn to adapt and live with it.
Q. Why are forecasts fundamental to engineering design?
A. Design, engineering design in particular, is about creating products for future use and enjoyment.
Picking up on the idea that good design provides good value for the money or effort expended, we want to shape and size our designs to what people will value. Good design thus implicitly relies on forecasts of what might eventually be needed or desired.
For example, good design will not create a "bridge to nowhere". The bridge itself might be structurally and technically sound. But if it serves no purpose, it is a waste of money and not good design.
Q. In Chapter 3, you state that "flexibility in design routinely improves performance by 25 percent or more." Can you give an example of how design flexibility can increase a project's value?
A. Design flexibility improves performance by allowing us to adapt products to future circumstances and thus to obtain good performance under a range of situations. A product without design flexibility, without the capacity to adapt, often leaves us with the wrong thing at the wrong time.
The book makes the point with the simple example of the design for a multi-story parking garage.
A standard design delivers a facility with a fixed number of levels and is likely to be too big at the beginning, and then too small when traffic builds up. However, a flexible design starts small and expands if and when needed. The flexible design thus avoids losses, and takes advantages of upside opportunities. It thus can provide a "win-win" product. Case after case show that the expected benefits from flexible design are truly significant.
Q. What are the different kinds of flexibility for a system and how do engineers determine which flexibilities will add the most value?
A. An obvious kind of flexibility concerns the size of a product – can we easily expand it if needed?
For example, can we double-deck a bridge to carry more traffic, as we did for the George Washington Bridge over the Hudson River?
Another flexibility concerns capability— can we upgrade it? An example is replacing a component with a new, better version, as aircraft designers routinely do when they modernize aircraft with more powerful engines or better computers.
A different flexibility enables changes in function — can we re-purpose a product to satisfy different needs, as we do when we transform office buildings into apartment houses?
Identifying which flexibility might add the most value is a challenge! The book discusses this process. As the value of flexibility lies in the capacity to adapt to the range of futures, we first need to focus on what the range of futures might be. We then need to analyze the possible flexibilities to identify those that add the most value. The choice of flexibility thus depends on the dynamics of the context. What is right for a high-tech product may not be best for conventional construction. What is appropriate for Singapore might not be best for the United States.
Q. Though it's impossible to predict the future, what are the key factors that engineers should consider when designing a project or system?
A. The key idea is that "the forecast is 'always wrong'"! Designers need to recognize the great likelihood of unforeseen developments that affect the performance of their products. Economic cycles alter demand, new technologies create new opportunities, and political events drive industries and markets. We need to be modest about our ability to specify requirements for a design. We thus need to recognize the great need for flexibility to adapt to future realities.
Q. Can you describe the common obstacles that engineers face when trying to implement design flexibility?
A. A most common obstacle is the refrain that "flexibility is nice, but we can't afford it".
This mantra comes from folks who think they know exactly what is needed, and who fail to recognize the ranges of conditions for their designs. It reflects a mental block that turns a blind eye to the reality of uncertainty about needs.
Q. What is the "Flaw of Averages"?
A. The "flaw of averages" is the mistake (the flaw) of believing that everything will average out (the "law" of averages). Designers make this mistake when they design products around "average" or "most likely" conditions — rather than design for the range of possible futures. When designers design around averages, they systematically estimate the value of their designs incorrectly, and make the wrong design choices.
The "flaw of averages" is a very simple and long-established mathematical fact. (It's also called "Jensen's law" and you can Google it for details.) Unfortunately, standard design processes fixed on a single set of requirements routinely fall into this error.
Q. What kinds of changes (if any) do you think we need to make in engineering education?
A. As regards design, the most important change is for educators to open our students' minds to the reality that forecasts of what is needed are imprecise at best. When we recognize this reality, we re-frame the design problem from meeting some fixed set of requirements to that of being able to adapt to meeting a range of requirements. The rest follows!