Digital versus Print: Energy Performance in the Selection and Use of Scholarly Journals

Advances in digital technology and the growth of information networks are revolutionizing human activity. The Internet has been championed as a new tool for environmental improvement. A life‐cycle energy analysis of digital libraries, a growing application of information technology, was conducted to test this premise. Life‐cycle models were compared for journal collections in digital and traditional formats. The basis for analysis was the amount of information in a typical scientific journal article (∼12 pages), which is equivalent to 0.97 hr of on‐screen reading time. Digital system elements such as servers, routers, laser printers, and computer workstations were modeled. Journal production, delivery, storage, binding, interlibrary loan, and photocopying were examined for the traditional system. Building‐related infrastructure, office paper, and personal transportation of the library patron were analyzed for both cases. In all, the study incorporated nearly 30 model elements, 90 input variables, and numerous fixed parameters. Five primary scenarios were constructed to consider increasing levels of complexity. Scenario 1 assumes only one reading per article (unit of analysis). Additional scenarios assume 1,000 readings and vary the following: laser printing, photocopying, and personal transportation. Energy consumed by the digital collection ranged between 4.10 and 216 MJ. The traditional system realized burdens from 0.55 to 525 MJ. Four significant effects were uncovered: (1) Energy consumption per unit was highly influenced by the number of readings per article. (2) Networking infrastructure by itself had a relatively small effect on total energy consumed by the digital system. (3) When personal transportation was considered, its effects tended to dominate. (4) The impact of making personal copies varied. Photocopying always increased energy consumption, whereas laser printing actually saved energy when it substituted for on‐screen reading.