Adam took the courageous dive into the world of blogging about us at Caltech living in “the gap.” I dare not commit to regular contributions as he admirably has, but have agreed to write a series of posts contrasting R&D for power and communication networks. This is the first in this series, and my first-ever blog post!
Smart grid is in vogue these days, for good reasons. As always, excitement garners people, ideas, and resources; but if not well-managed, can create disillusion that pushes the pendulum back. It’d be impossible to predict how the current resurgence of interest in power systems R&D will play out, but the confluence of powerful forces will likely (and hopefully) drive dramatic advances in the coming decades. We plan to chat about these in this space over the coming months.
Starting at the beginning
Despite their important differences, it helps to compare the evolution of two of the world’s biggest and most complex infrastructures. The communication network (telephone by Alexander Bell) and the power network (alternating-current transmission by Nikola Tesla) both started about 130 years ago. Their developments bore many similarities — both started as regulated monopolies, both were vertically integrated and tightly optimized, both were designed to deliver a single commodity extremely reliably, both went through rapid growth through two world wars and post war, and both even started deregulation around the same time in the 1980s-90s — until about 20 years ago, due to a little experiment that DARPA started in 1969, when the development of the communication network took a dramatic turn; more on this later.
Bell demonstrated in 1876 the first ever wired conversation over a two-mile wire between Cambridge and Boston, and in 1915 the first continental telephone call between New York and San Francisco. Tesla presented in 1888 “A new system of alternate current, motors, and transformers” to the American Institute of Electrical Engineers where he described his vision of a poly-phased infrastructure for long distance transfer of electric power (different from the direct-current network that Thomas Edison advocated and started to build). Interestingly Karl Benz was issued a patent of his internal combustion engine around the same time (1886) that led to the first production automobile. The late 19th century certainly seemed like a time of constructive disruption. 130 years on, now is such a time for us — if we can excite a new generation of students to take on the biggest challenges of our time, energy and environment.
It has been said that Bell would not recognize today’s communication networks, whereas most things would look familiar to Edison (and Tesla) if they were to visit us. Our local utility company still has transformers that were installed pre WWII, and they worked fine! Arguably, Daimler and Benz (both held patents on internal combustion engines) might recognize the basic structure of today’s cars — even though there has been dramatic progress in the control and robustness of automobiles, most of the complexity is highly structured and hidden. Both our power and transportation systems are about to change beyond recognition.
The importance of layered architecture
What has changed that has made today’s communication networks unrecognizable to Bell? Too many. But if there is a single most critical change, it would have to be the layered architecture of the Internet. The importance of architecture is widely recognized but rarely formalized. Internet has not only revolutionized communication networks, completely replacing the original telephony architecture and restructuring the landscapes of multiple industries, it has also developed into a common platform for innovations with impacts far beyond communications. Architectural design, however, remains an art. The state-of-the-art still relies too much on trial and error. Good architecture has been easy to recognize after the fact but elusive to forward-engineer or predict. Yet, over the next few decades, as we adapt the Internet to address new challenges, transform the nation’s power grids, and tinker with biological networks, architectural design must be more than an accident. We must develop a mathematical underpinning of network architecture and systematic methods to develop and evaluate design choices and algorithms, a holy grail that demands sustained effort.
John Doyle thinks most deeply and broadly about architecture across a wide variety of complex systems (technological, biological, social). Hopefully we can convince him to write a post on this in the future.
Will there be a layered architecture for power systems?
Any complex system is modularized and therefore one can almost always create some layer interpretation of the structure. Take frequency control as an example. Frequency control is mainly implemented on the generation side in today’s power network. It maintains the system frequency tightly around its nominal value (60Hz in the US and 50Hz in Europe) when load and generation fluctuate randomly. It consists of (at least) three mechanisms that work at different timescales from seconds to minutes, in concert. One can therefore interpret the primary, secondary, and tertiary frequency control as a layered architecture, layered along the time dimension. For another example, today’s grid consists mainly of bulk generators, connected to high-voltage long-distance transmission networks, that interface with low-voltage short-distance distribution networks, that eventually connect the end users, residential, commercial, and industrial. Different pieces are not only separated geographically but also provide different functions. We can also layer along organizational boundaries. One can therefore rightly interpret this as a layered architecture, layered along the space, functionality, as well as organization dimensions. These perspectives are all useful, but miss the point.
The traditional telephone network was also modularized in time, space, functionality and organization as in today’s power network. But it has a drastically different architecture from the Internet. The thin waist of the TCP/IP layered architecture makes the network not only extremely robust to failures and attacks (one of the original motivations of the ARPAnet, the precursor to the Internet, was the survival of the network in a nuclear attack), but more importantly, robust to technological evolutions above and below the waist. This allows different parts of the network to be designed independently, deployed asynchronously, and evolve rapidly.
This drastically speeds up innovations and their deployment, draws in talents and investment, spurs the explosion of startups, disequilibriates and disrupts, …, and the rest is history.
I’d be very interested in seeing statistics on the numbers, types, sizes, jobs, and capitalizations of networking startups before and after 1990 (any pointer will be greatly appreciated!). I think it will be a telling indicator on how the architectural change, a seemingly unimportant engineering tinkering, has had spectacular and positive impacts far beyond engineering. Some of the largest companies today did not exist 15 years ago. Some of the social movements today would not have been possible 15 years ago. The cultural, social, educational, political, economic, industrial and scientific landscapes have been dramatically altered, mostly (but not all!) for the better. I feel fortunate to be part of the research community that bears direct witness to this transformation.
The power network will (hopefully) undergo a similar architectural transformation in the coming decades that the communication network has gone through in the last two, with far-reaching and long-lasting impacts.
What is so difficult about a layered architecture for the power network? I’ll pick this up next week.