NSF Stories

In Industrial Manufacturing, Efficiency Falls as Technology Advances

A comprehensive study of old and new manufacturing processes, from machining metal to making carbon nanofibers, shows that the more advanced technologies are less efficient in their use of energy and materials per kilogram of output

A century ago, industrial manufacturing was dominated by large-scale, energy-hungry processes such as smelting ores and machining metals. Today, the industrial landscape features an array of novel techniques operating on much smaller scales, making computer chips, carbon nanofibers and other sophisticated products. That evolution toward more technologically advanced manufacturing has happened, a new study shows, at the cost of a sharp decrease in efficiency. The most hi-tech processes can take as much as a million times more energy and materials to make a given quantity of finished product than traditional industrial methods--a trend that may hinder efforts to build a more energy-conscious industrial economy using cutting-edge materials and technology.

Any manufacturing process uses energy to transform starting materials into a desired product. But some fraction of the starting materials will generally go to waste, and energy is not always used as efficiently as it might be. To better understand these losses, Timothy Gutowski of the Massachusetts Institute of Technology in Cambridge, Mass., and his colleagues have developed a framework, based on the laws of thermodynamics, that keeps track of both the energetic, as well as the physical or chemical, transformations of materials as they pass through the steps of an industrial process.

The researchers surveyed 20 manufacturing techniques. Traditional processes included melting and casting metal, grinding and machining, and injection molding of plastics, along with more recent innovations such as the use of  lasers and water jets to shape  materials. Rounding out the list of techniques were the wide range of processes used in the microelectronics sector, such as chemical vapor deposition and sputtering, as well as carbon nanofiber production.

With the exception of the methods that involved melting metal, the overall power requirements for each process were surprisingly similar, ranging from about 5 to 50 kilowatts of electricity. On the other hand, the amounts of material processed varied enormously, ranging from hundreds of kilograms per hour or more for the older processes to just a few milligrams per hour for two of the most novel techniques. A striking trend emerged: as processes become more technologically sophisticated, they tend to manipulate smaller and smaller quantities of material at slower rates, but since power consumption per process has stayed about the same, the amount of energy needed to generate a given quantity of finished product has been growing fast.

An inefficient process operating on a tiny amount of material may not be very important when set against the entirety of industrial energy use, Gutowski and his colleagues observe. They point out, however, that the world now produces more than 20,000 tons of electronics-grade silicon per year, consuming about 50 billion kilowatt hours of electricity in the process. Production of carbon fibers, which are often proposed for large scale applications, is in the same ballpark in terms of energy used per kilogram of product. Both industries involve processes that fall in the middle and lower range of efficiencies found by the MIT researchers, so that the lack of attention to making better use of energy could have significant consequences.

Modern industrial techniques often require elaborate materials and procedures that exact energetic and material costs but don't get directly incorporated into the product. For example, highly reactive gases may be used to clean production equipment for silicon chips in preparation for other steps, and those gases may have to be chemically treated after use for safety or pollution-control reasons. These ancillary but essential elements of a hi-tech manufacturing process can reduce enormously its "degree of perfection"--the ratio of the thermodynamic value of the product to the thermodynamic value of everything that was needed to make it. The more ideal a process, the closer that ratio will be to one. In reality, processes can vary significantly in their degree of perfection. For example, the MIT team calculated that an electric furnace that melts scrap and other iron to produce a refined metal output can have a degree of perfection of 0.79. By contrast, a chemical vapor deposition method used by the semiconductor industry to produce thin layers of silicon dioxide can have a degree of perfection of not quite four millionths.

In designing processes that turn expensive materials into small quantities of hi-tech products, manufacturers have focused their attention on a variety of issues such as size and quality, but "haven't had strong incentives to reduce energy consumption," Gutowski says. But that may be changing. Manufacturers of solar panels, for example, are well aware of the energy-intensive nature of the processes they use, and figure it into estimates of the pay-back time--in energy as well as money--of the devices they make. To be truly green, in other words, a solar panel must deliver substantially more energy over its lifetime than was consumed in creating it.

That kind of thinking hasn't yet taken hold in other areas, such as nanomaterials, Gutowski adds, where the energy costs of manufacturing are not widely known. As more applications of such materials gain attention, especially in the context of "green" technology, "there will have to be a serious conversation about dealing with energy costs," he noted. By providing a comprehensive analytical model that can account for both material and energy use in industrial manufacturing, the MIT study provides the language in which that conversation can be conducted.