Quanta Technology Blog
Technological Forecasting & Distribution Reliability
Posted on: Jul 15, 2015
We received a good number of emails about last month’s blog on technology forecasting, so this month’s discussion will delve a bit deeper into some of the ways technology can be projected and anticipated. Readers wanting more detail and depth than I can go into here, can check out the latest edition of J. P. Martino’s Technological Forecasting for Decision Making. I learned the basic principles from the first edition more than thirty years ago.
The S Curve Rules. The trend of practical development of a technology almost always follows a Gompertz curve, often called an “S” curve, as discussed last month (Figure 1). There are numerous reasons why, many fairly obvious and a few not. In nearly all cases, whether it is the top speed of jet aircraft, the SEER of residential air conditioners, or the net operating efficiency of fuel cells, an S curve in some form describes the trend of improvement of a particular technology over time.
Figure 1: Development of a device or system almost always improves in a non-linear way over time, although the shape (it may not be symmetrical) and pace (how quickly change happens) of the curve may differ a great deal depending on specifics.
Materials and Variability. The two major anticipatable - and therefore forecastable - areas in a technology are its use of materials and any designed-in variability. Both are, in their own way, analyzed and projected into the future.
The key material properties on which a device, or system’s performance depends, are improved over time. Temperature-tolerance is improved by development of new alloys; ways are found to remove micro-contaminants that limit efficiency; nano-arrangement of structure is pushed to a finer level to provide faster or denser operation; the improving purity and nano-positioning of ions in batteries, and the use of better alloys and newer materials (hardened aluminum for steel in light trucks, ceramics replacing metal in heat pumps) are some actual examples.
Variability is the traditional means of “having your cake and eating it too” with regard to performance of a device or system. Its use goes back centuries to the first drawbridges and canal locks. It remains one of the major ways performance of engineered devices is improved over time. The classic example is the number of speeds in the automatic transmission of cars. A car with a two-speed transmission can operate its engine most efficiently at only two speeds. In daily use, it will often be operating at noticeably non-optimal engine speeds. A three-speed transmission will allow a car to operate closer to optimum much more often. It gets better fuel economy, and perhaps also accelerates faster and provides quieter and smoother operation than the car with the two-speed transmission. A four-speed transmission will do even better, and so forth. This is why, over the past seventy years, automatic transmissions in cars have steadily increased from two speeds to as many as nine in some cars today.
Variability has been used for decades in power systems. Voltage regulators and switched capacitors are two examples. Another is feeder switching. Early feeder systems had no switching and were not laid out to make use of switching. Modern feeder systems have a good number of switches distributed throughout a system designed to have alternate pathways. Switching time has also improved, from more than half an hour to operate switches in a coordinated manner to less than a second in some cases, so that that increased reconfiguration capability can be applied quickly and conveniently. As a result, the reliability of radial feeder systems has improved.
Go All The Way to Set the Limit. A cornerstone of technological forecasting, and usually the first step, is to identify the upper asymptotic limit of the S curve (dashed line in Figure 1) by assuming perfect materials are developed and used, and that all possible variability is built into the device or system. For example, the ultimate performance of lithium-ion batteries can be estimated by assuming absolute purity of materials along with arrangement in an optimal structure (from the standpoint of energy storage density) at the atomic level, etc. Such assumptions lead to a conclusion that such battery technology can potentially store up to three times as much energy per packaged pound of battery as today. Similarly, one can determine the fuel economy and other advantages that would be delivered by an automatic transmission with an infinite number of gears, or the reliability that a feeder system with a switch at every node would have. In this way, the asymptotic limit of the S curve is first determined.
The Rate of Development. The shape and slope of the S curve is then forecast through analysis of the development or production pace of the technology. Past performance in a field is the best indicator of this, and so technological forecasting is much better when applied to a technology with some history to it. For example, over the past sixty years it has taken the auto industry about eight years to increment the number of gears in an automatic transmission by one. By contrast, battery development cycles seem to take about five years now, while successive generations of fusion-power reactor designs take about twelve to fifteen years each. From these analysis and detailed studies that build on that analysis, one concludes that it may not be worth pursuing ten- and twelve-speed transmissions, that it will be another fifteen years before battery technology based on lithium reaches full maturity, and that fusion power may not be a reality until late in this century or the first quarter of the next.
This step in technological forecasting also often uncovers caveats about certain types of technologies that help forecast where some development paths may be a dead end. For example, it is possible to build automatic transmissions that are continuously variable – that have an infinite number of gear ratios. However, theoretical studies can show that purely mechanical continuously-variable transmissions have energy losses (internal friction) that limit their fuel economy to about that of five-speed transmissions, and electronic variable-speed transmissions can’t do much better. Thus, the continued focus by manufacturers on adding more discrete gear ratios to what are otherwise traditional transmission designs.
Figure 2: A new disruptive technology (red) may intervene to extend a technology that was near its peak, as occurred when jet engines replaced propellers in airplanes.
Watch for Disruptive Technologies. Figure 2 shows a double S curve – a new technology provides the performance in a different way with a much higher asymptote. A classic example is the top speeds of military aircraft. Propeller-driven aircraft are limited to about 550 mph due to the increasing inefficiency of propellers as their tips approach the speed of sound. Jet engines provided an increase of the asymptote to four to five times that – limited not by propulsive physical laws but the strength of materials of the airframe.
Similarly with regard to automatic transmissions, hybrid drive trains blur the line between engine and transmission. Taking that into account, hybridization of cars moves the transmission-dependent fuel economy asymptote up by about 60% from where it would otherwise be. With regard to batteries, newer battery technologies may eclipse lithium, and more generally in any field, a new basic technology or approach may “change the rules.”
Follow the Trend and Physical Laws Mindlessly. Like quantum mechanics, technology forecasting is a field where it is best to apply no imagination or judgment. It is best applied without regard to whether one can picture its implications. In projecting the continued development of lithium-based batteries, have only the vaguest idea how anyone is going to figure out how to arrange and keep lithium atoms and ions in their optimal structure and package them efficiently once the materials are prepared. But the rules say there are no natural physical limits at hand to stop someone from figuring out how to do it; however, past experience in all fields proves that innovation eventually finds a way. Thus, the best forecasts ignore the fact that one’s imagination may not be able to fathom how it will be done, but stick to the analysis and trends that say it will be done.
Intervention by Other Metrics. A challenge for the technological forecaster is to recognize when another metric may intervene to make the trend being forecast irrelevant or less important, so that it no longer drives the trend of a technology or device’s development. For example, the energy efficiency of home clothes washers has improved little over the past two decades, and yet it is possible and practical to build automatic washing machines that use only about half the energy of the best available today. The reason energy usage has not been further improved is that minimization of water usage is considered more critical and that, often fueled by use of energy rather than water, is driving clothes washer technology right now.
Technology-Specific knowledge is Essential. The foregoing discussion has probably led many readers to think that there are so many ifs, ands, and buts involved in the steps discussed above that useful technological forecasting is basically impossible. It’s not quite that bad. As far as being impossible, it is not. But what has proven to be the case is that detailed, industry-specific knowledge is always needed in order to sort out all the details and do a good job of forecasting.
For this reason, one finds no really successful general technological forecasters – no savants who forecast battery technology this month, pharmaceutical trends the next, deep space probe robotic capability a few months later, and undersea drilling technology as their final project of the year. Instead, technological forecasting in almost always done by people who understand the ins and outs of the technology and its use, and have learned technological forecasting methodology as they need it. I think that is why it continues to be a less-than-well-recognized field.
Even its acknowledged experts, like Joseph Martino, work mostly in just one field (aircraft technology). I’ve attended conferences, seminars, and workshops on technological forecasting, and met people who apply it in other industries and fields, and people who teach its principals at universities, but I have never applied it out of the power field. I know I don’t have the skills, and I doubt anyone outside of the power field does.
Figure 3: Expected System Average Interruption Duration Index, or SAIDI, of a particular electric distribution system in the south-central United States, as automation, distributed resources, and control become more affordable and useable over time.
Put all the foregoing together in a detailed study and projection of a technology and quite interesting forecasts are often produced. Figure 3 shows a forecast recently done for a local delivery utility in the central United States. It plots the economically justifiable SAIDI performance that can be expected on this utility’s distribution system as feeder automation systems and various distributed technologies improve and it becomes affordable to add more to the system in the future. Readers should note the quantitative results shown are very system specific – they depend on a host of details specific to this utility, including exactly how it computes SAIDI, how it defines “economically justifiable”, the type and condition of its system, its load characteristics, and a myriad other criteria. But qualitatively, many utilities would see something similar. Utilities that saw different results, and there would be a few, would perhaps find the results just as interesting, but leading to different decisions and strategies. Regardless, the plot shown has a number of interesting features and proves useful for many aspects of the utility’s planning for its future. That is, ultimately, what technology forecasting is all about.
Posted by Lee Willis on Aug 20, 2015
Distributed Secondary Networks (DSNs) are ubiquitous in the downtown areas of North American cities. They serve the vast majority of central high-rise business districts and... Read more
Posted by Lee Willis on Jul 15, 2015
We received a good number of emails about last month’s blog on technology forecasting, so this month’s discussion will delve a bit deeper into some of the ways... Read more
Posted by Lee Willis on Jun 16, 2015
In early 1953, my grandfather bought a new Cadillac Coupe DeVille. I was a young child at the time and thought that it was the most wondrous thing I had ever seen, full of magical... Read more
Posted by Lee Willis on May 28, 2015
Twenty years ago, I did an interesting project for a friend who had just become COO of a new and rapidly growing cable TV company. He asked me to look at how his cable and... Read more
Posted by Lee Willis on Apr 22, 2015
Traditional electric utilities that do not adapt to a “private microgrid” world could be looking at eventually losing up to half their customer base.... Read more
Posted by Lee Willis, Damir Novosel on Mar 24, 2015
Renewable energy, distributed resources, smart equipment, microgrids, customer-centric energy plans and a host of other new concepts and technologies are changing our industry... Read more
Posted by Lee Willis on Feb 23, 2015
Writers of adventure and espionage fiction have always stayed a step ahead of the real world, but only a step. The best, or at least the most successful of them, seem to have a... Read more
Posted by Lee Willis on Jan 20, 2015
We are far enough into the current century that our present technology is noticeably better than it was at its beginning. Our latest automobiles are cleaner, safer and get better... Read more
Posted by on Nov 21, 2014
Just what exactly is going on down there? On July 14, 2008, at 8:54 a.m., the city of Vancouver, BC made national and international headlines after a... Read more
Posted by Lee Willis on Oct 16, 2014
For most of the past one hundred years, new technologies have entered the power industry "at the top", near the generation level, and gradually worked their way down to the... Read more