l:ffiEW Vo lume 21 Number Three 1988

Oak Ridge National Laboratory is a multiprogram, multipurpose laboratory that conducts research in the physical, chemical, and life sciences; in fusion, fission, and fossil energy; and in energy conservation and other energy-related technologies.

I ON THE COVER I

Th is computer graphic (produced by Pau l T. Editor Williams of the Carolyn Krause Martin Marietta Energy Systems, Associate Editor Inc., Computing Luci Bell and Tele­ communications Consulting Editor Division) is a Alex Zucker "finite elements" model developed Designer for vibrational analysis of part of Vick ie Conner a scanning tunneling Technical Editing microscope being Lydia Corrill built at ORNL. Color variations Photography, indicate the Graph ic Arts, and degree of surface Pri nti ng and Duplicating change (see page departments 9 for the full story). A special section of this issue is devoted to the applications of supercomputers, particularly the local Cray X-MP, in research and in visuali zing scientific resu Its at ORNL.

The Oak Ridge Na riona/ Lahorarory Re1·iew is publ ished quarterly and distribut ed to employees and others associated wi th ORNL. The address of the editorial offi ce is Buildin g 4500-South, M.S. 6 144. Oak Ridge Nati onal Laboratory, P.O. Box 2008. Oak Ridge, TN 37831-6144. Telephone: inte rnal, 4-7183 or 4-6974; commercial, (6 15) 574-7183; or (FfS) 624-7 183.

O RNL is operated by Marti n Mari etta Energy Systems. Inc .. for th e Depart ment of Energy un der Contract No. DE-AC05-840R2 1400

lSSN 004R-1262 Features

SUPERCOMPUTERS IN SCIENTIFIC RESEARCH 2 H. Richard Hicks

ACCOMPANYING ARTICLES:

Using a Supercomputer for Theoretical C hemistry 10 Supercomputing and Reactor Safety 12 Supercomputers in Materials Science Theory 14 Fusion Calculations on Cray Supercomputers 17 Theoretical Atomic and Nuclear Physics 19 Parallel Computers for Efficient Problem Solving 21

Energy for Development: ORNL Returns to the Third World 24 Tom Wilbanks and Sherry Wright

Designer Steels for Advanced Energy Applications 42 Carolyn Krause

New DNA Stain Aids Cell Studies so Luci Bell

Departments

Take a Number 41

Awards and Appointments 54

Books: Chaos, by James Gleick, rev iewed by Woody Gove 58

R&D Updates 62

Alvin Trivelpiece: ORNL's new director Two HHIRF devices operating Tritium pell et injector fo r fu sion demonstrated DOE Superconductivity Pilot Center located at O RNL

Technical Highlights 65

Hood Ri ve r Conservation Project an energy- aving success Decline in tree growth li nked to aluminum "freed" from soil by acid rai n

Technology Transfer 68

Vall ey-Todeco li censed to make ai rcraft fasteners from OR L all oy Metallamics to man ufac tu re nickel al umi nide products Supercomputers in Scientific • This image is a Research supercomputer simulation of an By H. Richard Hicks astrophysical explosion between two colliding neutron stars. The image upercomputers are powerful machines determining the actua l speed ach ieved. Compiler (originally in color) was created by used to predict the weather, model the optimization refers to the degree to which a § climate, design buildings and veh icles, computer program is translated into machine Donna J. Cox , artist and color and locate oi l reserve . At Oak Ri dge ational instructions that permit efficient execution of the software designer, Laboratory they are u ·ed for sc ientific research in program. and Charles R. areas such as fusion energy, chemistry. physics. Because supercomputers are near the leading Evans Ill, astro­ metallurgical theory. and reactor simulat ion . edge of technology. some prototypes or research physicist. Cox The term supercomputer simply mean one of machines are not widely available to end users. used this in a the most powerful computers avai lab le at any These machine are of great importance but are sem inar at ORNL given time. Thi s working definition avoids any not generally relevant to the typical supercomputer in January 1988 to controversy about ''How fast is fast?" and reflects user. illustrate the use the expectation that more powerful supercomput­ When more "capabi lity" (ability to proce of supercomputers in scientific data ers will evolve. bigger jobs in the same time) is needed, then a visualizati on . Supercomputer al o usually means a machine faster computer may be th e only answe r. When that i not embedded. Embedded computer are more "capacity .. (ability to process more job in those built into device such as car. or rockets the same time) is required, it can be ach ieved and having fairly specific roles. Usually . . uper­ either by a faster computer or by simply adding computers are th ough t of as being gene ral­ more computers. As one author put it. the differ­ purpose mach ines. although some have been ence between a faster computer and a host of designed to solve narrow classes of problems. computers working together is like that between a The term .. powerful" refers to the arithmetic or Saturn-Y rocket and a fleet of automobiles. They logical processing speed of the computer. Super­ may have the same total power. but the rocket and computers normally have other re ources. such as the cars are optimized for different purposes. memory size. communication speed. and storage capacity that are commensurate with the process­ Vector Processing ing speed, but these factors are usually not considered to be primary measures of power. In the early 1970s. the computer industry Sometimes the maximum hardware speed. in rea lized th at the speed of electronic sw itch ing in millions of instructions per second (MIPS) or in computers was not increasing as rapidly as it had millions of floating-point operations per econd in the previous decade. It was also recognized that (MFLOPS). is quoted as a measure of power. A floating-point arithmetic ope rations generally took floating point operation is an arithmetic operation a lot longer to complete th an most oth er opera­ between two real numbers. However. the speed of tions. Th is recognition prompted an architectural computation on most modern supercomputers advance called vectorization. bear little re semblance to this maximum number. This is how it works. Suppose that five clock because programming techniques and compi ler periods are used to multiply two numbers together. optimization play an important role in It is possible to des ign a multiplier in the form of a

2 Oak Ridge National Laboratory REVIEW five-segment .. pipeline." The first segment is busy extreme are computers that store all data in a during the first clock period, the second during the common memory that can be accessed by all second clock period. and so forth. This arrange­ processors. Any data to be passed between proc­ ment does not provide any advantage for a single essors must go through this memory. At the other operation. However, consider the case in which a extreme are computers in which each processor sequence of pairs must be multiplied to produce a has a local memory. A communications network sequence of products. If all of the pairs are known is used to move data between processors. In local­ in advance (that is, if none of them depend on any memory multiprocessors (sometimes call ed of the products), then, when the pairs are fed distributed-memory machines), the topology of through the pipelined multiplier, the first product the internal communications network is the key. will appear after five clock periods. After that, one Maximum communication is provided by directly product will appear per clock connecting each processor to every other proces- period because a new operation sor. Each message can then begins each clock period. These go directly to its destination: sequences (of either operands or but, for computers having a an wers) are called vectors. For large number of processors, vectors much longer than five. the number of connections is one approaches the ideal of one very large. The minimum result per cycle-five times the communication is provided by scalar speed. a ring topology, which may The computer require many hops for some drawing of the Parallel messages. (A hop is the Gray X-MP was done by Gail Finch Processing number of time a message must be sent from processor to of the Energy Systems Comput­ In a limited sen ·e. processor to reach ing and Telecom­ vectorization all ows its final destina­ munications several operations to tion.) Another Division . be in progress at once. natural topology is In contrast, the term a two-dimensional parallel processing is rectangular grid . appli ed to computers Perhaps the most that have several fairly elegant, and the best complete processing units. compromise between (See article on parallel computing; minimizing connections on page 21.) When these units are and minimizing hops, is the all owed to process different in struction st reams hypercube topology, which requires the number and different data streams simultaneously. the of processors to be a power of two, 2". The total machine is classified as a multiple-instruction, number of connections between processors in this 1 multiple-data (MIMD) machine. In contrast, the configuration is d x 2"- • and the maximum older. standard. serial processor would be a single­ number of hops a message will take is d. instruction, single-data (SISD) machine. and a By far the largest installed base of supercom­ machine that operates in parallel on arrays of data puters is of those manufactured by Cray usi ng a single stream of instructions would be a Research . Inc. In the early 1970s. Seymour Cray, si ngle-in struction, multiple-data (SIMD) machine. the developer of th e CDC 7600. left Control Data Even these categories barely scratch the surface Corporation to found his own company. Cray in describing this rich class of computers. The Research. Inc .. which focuses exclusively on memory's organization plays a very important role building supercomputers. In the mid-1970s, this in the operation of a parallel computer. At one company marketed the Cray- 1, which was about a

Number Three 1988 3 SUPERCOMPUTERS IN SCIENTIFIC RESEARCH

increased capabili ty. In fact, although maximum theoreti cal processing speed has inc reased, no dramati c increases in th e actual processing speed fo r most typical calcul ati ons here occurred since the introduction of the Cray- l. The primary difference between the Cray X- MP and the Cray-2 i in memory ize. The X-MP offers up to 16 megawords (memory units consisting of one million 64-bit words). The Cray-2 offers up to 256 mega word s of memory, all owin g calcul ati ons using large amounts of data 8-PROCESSOR 8-PROCESSOR to be executed more efficiently. ETA, a spin-off RING PERIODIC 2d GRID company of Control Data, has produced super­ computers hav ing greater theoreti cal speed than the Cray machines, but to achieve this greater speed, the calculat ion mu t use very long vectors. Recentl y, IBM and a number of Japanese companies have entered the supercomputing arena. A number of ''minisupercomputer'' vendors now claim I 0 to 40% of th e speed of a Cray-1 for. at most, a few hundred thousand doll ars. (Cray computers usuall y cost several milli on doll ars.) Hence, the day of Cray's virtual monopoly in the supercomputer market are past. 8-PROCESSOR 16-PROCESSOR Although the large commercial vendors have TOTALLY CONNECTED HYPERCUBE concentrated on shared-memory machines hav ing a sma ll number of fast processors, uni versities and a few small companies have foc used on hyper­ cubes and other computers hav ing large numbers factor of 4 faster th an the CDC 7600. the premier of proce. sors. One limitati on of hav ing multiple supercomputer at the time. The Cray- 1 achieved processor i th e conflict over access to a shared Various configura­ thi s speed because of its vector architecture. memory. Hence, these machines usuall y have a tions for connect­ Although Seymour Cray speciali zes in hardware local memory for each processor and a means to ing the processors design and packaging. Cray Research also communi cate between processors. Major efforts in different types prov ides it cu ·tomers wi th the software to all ow are under way at OR L and elsewhere to design of multiprocessor standard FORTRA programs to take advantage algorithm th at wil l execute optimall y in this supercomputers. of thi s vector capabili ty. environment and to develop tools to help the Cray has also developed and marketed the Cray programmer utili ze the computer's inherent X-MP, Cray-2, and Cray Y -MP computers, para ll el ism. whi ch have prov ided incrementa ll y greater processor speed and whi ch all ow up to eight proc­ Who Needs Supercomputers? essors per computer. The abil ity to harness these processors fo r a single job requires effort on the Several year. ago, the ati onal Science Founda­ part of the programmer, becau e each proce. sor ti on (N SF) began to recognize that U.S. uni versity usuall y works on a separate job. Thus. more professor and students needed supercomputer processors provide increased capac it y, but onl y access and training. To ac hi eve thi s goal, th e NSF with programming effo rt does thi s prov ide established five supercomputer centers, whi ch are

4 Oak Ridge National Laboratory REVIEW now operational: The Nati onal Center for Super­ computer Applications (U niversity of Illinois), the San Diego Supercomputer Center (U ni versity of California, San Diego), th e John Yon Neumann Center at Princeton, the Pittsburgh Supercomputer Center, and the Cornell National Supercom puter Facility. They employ Cray, ETA. IBM, and Floating Point System s (FPS) products. Although thi s initiative has had the undesirable short-term effect of making programmers for supercompute r systems somewhat scarce. a generation of students and scientists is receiving valuable training. Some 6000 researchers around the co untry have shared com puter time at these centers. Th e NSF plans, as the next step, to establish a hi gher-speed commu­ ni cation network between th ese centers. U.S. industry is now beginning to recognize the importance of supercomputers for performing geologic calculations, engineering design, hi gh­ quality an imation, and even large, mundane business tasks. Thus. the commercial mark et for superc omputers is also growi ng. Historically , the DOE defense laboratories, Oak Ridge's Cray X-MP Lawrence Li ve rmore and Los Alamos. ha ve led The Gray X-MP the way in purchasing supercomputers and in The Cray X-MP, located at the Oak Ridge (shown in the developing software to run them . As supercomput­ Gaseous Diffusion Plant (ORG DP). was acq uired background) is ers become more widely used, vendors will be to handle uranium enrichment calcul ations in a controlled by researchers like able to provide more of the necessary software. classified envi ronment. Just as the machine was Richard Hicks, Today . using all the capabi lit ies of highl y obtai ned in 1985. that wo rk disappeared because working at an parallel machines such as hypercubes cannot be the plant was pl aced on standby. and th e gas­ ord inary computer done without in curring considerable added centrifuge enrichmen t projec t was terminated . terminal. programming costs. To do it economically The computer was retai ned, however. to satisfy will req uire software adva nces and , probably. other computing needs wit hin Martin Marietta programming languages that are desi gned to Energy Systems. Inc .. and is now operated for in corporate parallelism. Some studies suggest classified use du ring part of each week and un ­ that parallel computers will not be used as class ified use for the remainder. Most of the un­ general-purpose computers within the next ten classified tim e is used for ORNL projects. wh il e years because of the programming costs. the cl ass ified time is used primari ly by ORGDP Increasi ngly, supercomputers wi ll be considered and th e Oak Ridge Y-12 Plant. When the Cra y is merely another reso urce th at can be accessed un classifi ed. most Energy Systems employees from a personal workstati on. With th e advent of may access it. reli ab le, hi gh-bandwidth telecommunications , it At ORNL. thi s Cray and other superc omputers is possible fo r workstations to route work invi sibl y are used fo r th e simulation of complex physical to the most effective reso urces fo r executi on. systems and fo r large engi neeri ng studi es to Work-stations are now starting to provide rea l­ provide numerical results that augment theoreti­ time anim ati on of simul ations produced by super­ cal. heuristic. and probabilistic approac hes. In thi s computers, uch as the operation of a reactor or context. th e supercomputer is a useful tool for the ae rodynamics of an airpl ane. problem solvin g. The leve l of resolution (detai l)

Number Three 1988 5 SUPERCOMPUTERS IN SCIENTIFIC RESEARCH

progress has been made in the effective use of parallel architec­ tures for certain etas es of An arrangement of scientific computer chips problems. and microcircu its typical of the Future of electronics that drive a super­ Super­ computer. computers Even if the next­ generation super­ computer is only sli ghtly larger than today' personal com­ of the sol ution is usually limited by the power of puter, it is likely to cost million of dollar . But. as the computer available. Doubling the linear the desktop power available for thousands of resoluti on in a three-dimensional imulation. for dollars increa es dramatically, so will the power example, can require a factor-of-eight increa e in ava il able in a supercomputer increase. Many of computer run time. the problems being solved today on supercomput­ Ten years ago, the validity of numerical simula­ er will be studied economically on much fa ter tions was not widely accepted in the scientific machines. Today ' computerized physical simula­ community. Because of verified ucces e during tions have by no means ex hausted all possibilities. the last decade and the money saved by perform­ They incorporate numerous assumptions that can ing computer, rather than laboratory, experi­ be refined, yielding further understanding. The ments, numerical simulations in many fields are study of nonlinear phenomena, geometrically consi dered a acceptable as other theoretical and complex systems. and problems involving diverse laboratory techniques. (See the fo ll owing articles time or space scales, real-time animation. and a for orne detailed examples of supercomputer high degree of dimensionality will strain the capa­ uses by ORNL researchers.) bilitie of supercomputers for orne time into the Even before in. tallation of the local Cray, some future. Scientific problems that require even more ORNL researchers used remote supercomputer , computer power than can be delivered will persist. primarily tho eat the defense laboratories. For If software can be developed to facilitate the example. the researchers in ORNL's Fusion effective use of many parallel processors, then the Energy Division have historically used the largest widespread use of massively parallel machines annual amount of computer time at the National could enormously increase the effective comput­ Magnetic Fusion Energy Computer Center at ing capability available. This ass umption, how­ Livermore. ever, implies that compi lers will also recognize ORNL researchers have also been inve tigating and utilize parallelism. better ways to design and use computers. They Many researchers are also investigating the po - have explored the use of parallel processors in sibi lities for significantly increasing processor robotics and have optimized some numerical speed. Today's computer ch ips are based on algorithms for parallel proces or . Significant macroscopic electronic technology. Atomic- or

6 Oak Ridge National Laboratory REVIEW • View of the Cray X-MP's multiple connections.

molecular-based circuit elements could shorten the allowing the development of more-compact signal path lengths, and thus the cycle times, by computer architectures. If these developments several orders of magnitude. Some of these succeed, the trend toward heavy reliance on approaches may decrease heat buildup. thus parallel computing may reverse. =:I

Number Three 1988 7 Richard Hicks has used a Gray X-MP for studying the dynam ics of fusion plasmas.

Biographical Sketch H. Richard Hicks is the program coordinator for the Office of Scientific and Technical Computing in the Energy System · Computing and Telecommunications Division. He ha worked for this division since 1972. He holds a Ph.D. degree in physics from the University of lllinoi . An overall theme of hi s scientific work has been the study of nonlinear problems through computer simulation. For a decade, Hick s wo rk ed in the magnetohydrodynamics (MHO) project of ORNL ·s Fusion Energy Division. Project researchers, headed by Ben Carreras. pioneered techniques to perform effic ient MHO computat ion s for better under­ stand in g the dynamics of fusion plasmas. Hicks. formerly one of the largest u,er of supercomputer time in the world. developed the computer program RSF. In the early 1980s. this and other computing tools helped to lay the theoretical groundwork for the design of OR L 's Advanced Toroidal Facility, wh ich achieved its first plasma in early February 1988. Hicks has been involved in de. ign discussions fo r one of the earli est parallel computers. the ILLIAC IV , as well as projects to "parallelize" existin g computer programs for Cray supercomputers and hypercubes. Hicks is an editor of Joumal ofCompurarional Physics.

8 Oak Ridge National Laboratory REVIEW Cray X-MP Aids Visualization of STM Geometry

This "finite-element" model of part of a scanning tunneling microscope (shown in black-and-white here and in full color on the cover) aids the visualization of the surface geometry. A rapid change in color indicates high curvature, while uniform color represents a flat surface. The Cray X-MP was used in a vibrational analysis with the goal of controlling vibrations to ensure accuracy in the STM's imaging of materials with atomic-scale resolution. The STM is being built by John Noonan of ORNL's Solid State Division.

Number Three 1988 9 SUPERCOMPUTERS IN RESEARCH

Using a Supercomputer for Theoretical Chemistry

e cribing the motions withi n atoms molecular collisions or ab orption of a laser and molecules is a goal of chemistry photon, the motion becomes random, or chaotic. and physics. A detailed understanding This chaotic motion was unknown until 1964, Dof these "nonlinear dynamical systems," which when high-speed digital computers were first have received much attention since the Late used to simulate the motions of a set of objects 1970s, could provide insights into how believed to be governed by classical mechanics, they are affected by molecular such as the solar system. Before then, analytical "Chaos, col li sions, laser irradiation, and other methods always revealed only quasiperiodic perturbation . motion. Chaos, currently a hot topic in science, currently a Almost every molecular or atomic has been found in almost every dynamical system y tern has been found to exhibit two ranging from nuclei to planets. (See the book hot topic in very different types of motion. Ln the review of Chaos on page 62.) ground state, there is a stable, or During the past 15 years (I 0 of them at ORNL), science, has qua iperiodic, motion. However, if a I have studied quasiperiodic and chaotic behavior been found certain amount of excitation energy is in the infrared (IR) multiphoton dissociation of deposited in the system as a result of molecules, molecular colli sions, atomic dynamic in almost every 29.40 ~------, dynamical system rangtng• from nuclei to planets."

28.20 ~----.-,----.-,---.,-----.-,---,1-----,----,,,----.,-----.-,--~ 4.5 4.6 4 .7

This fractal region of the energy-transfer map for a ) helium atom colliding with a diatomic molecule of iodine (1 2 was produced by the author in collaboration with S. D. Gray and S. A. Rice of the University of Chicago.

10 Oak Ridge National Laboratory REVIEW in strong mag neti c fi elds, nuclear dy nami cs of the ca lcul ati on was needed to show th at thi s fun ction gamma- ray laser, and dynami cs of polymers (com­ has a dimension greater than th e normal Eucl id­ plex chemica l compounds consisting of repeating ean dimension of a line and is a res ul t of class i­ structural unit s). I have pionee red several of the call y chaoti c motion. methods used fo r th ese studies in coll aborati on In ves ti gating both th e phys ical and with M. L. Koszykowski of Sand ia Nati onal Labo­ chemi cal properties of po lymers is a ratori es and R. A. Marcus of the Californ ia re lated new initiati ve in the Chemi stry "Recently, In titute of Tec hnology. Di vi ion. As part of th at effort, we Most analyses of dy nami ca l systems require the have started to develop a computer we have use of a supercompu te r to solve large sets of model fo r the dynamics of large used the coupled, nonlinear diffe renti al equ ati ons or to po lymer molecul es. Our current goals diagonali ze large matri ces. Recentl y. we have are Cray used th e Cray X-MP at ORGDP to study chaos in 0 To conduct large-scale si mul a­ molec ul ar co IIi ions. Us ing I 0 h of computin g X-MP at tions of mac romolec ul ar moti on. time, we generated an energy- transfer map (s hown using molec ul ar dynamics on oppos in g page) of an atom co lli ding with a ORGDP models, and to study 0 To develop better meth ods to make these calcul ati ons ro utine. chaos in We pl an also to ex tend our under­ standing of non! in ear dy namics and molecular chaos in simpl e systems to th e more collisions." compli cated (man y-body) structu ra l probl ems of polymers. Our projects will in vesti gate th e detail ed process by whi ch a poly-e th ylene chain melts on a surface and the energy flow in polymers irradiated by a beam of IR light. We also pl an to calcul ate the various spectral properties of polymers and to study the Condi s crystal, a confo rm ati onall y disord ered ph ase in polymers di scovered by Bern ard Wun­ derli ch. a University of Tennessee-ORNL Di stin ­ gui shed Scienti st. Because of the large r number of atoms to be simul ated and, conseq uent ly, the much la rger number of numerical calcul ati ons Drawing of a polyethyl ene crystal. required. we are fortunate to have ava il abl e to us the capacity of th e Cray X-MP and th e support di atomi c molec ul e. Thi s map is an example of a from staff members of th e Energy Systems functi on ca ll ed a fractal (i.e .. any further magnifi­ Computing and Telecommunicati ons cati on, or finer grid . of th e map wo uld show a Di vision.-Don W. Noid. Chemislry Dirision similar set of scattered points). The ex tensive Cray

Number Three 1988 II SUPERCOMPUTERS IN RESEARCH Supercomputing and Reactor Safety

I 00 commercial nuclear power fission products. These reaction products would nts currently operating in the United heat and pressurize the primary containment, tares generate about 17 % of the total which could ultimately fail, releasing fission electricity produced annually. Forty of the e products, aerosols, steam. and combustible gases plants employ the boiling water reactor (BWR) (cumulatively referred to as '"blowdown"') into the design. Th~:~s , the continued surrounding reactor building or secondary contain­ safe operation of BWR pl ants ment. The aero ols and fission products would is a matter of great national im­ migrate through the reactor building, driven by "The scope of portance and public concern. primary containment blowdown and combustible phenomena Since 1980, OR L ha made gas deflagrations (combustion of hydrogen and significant contributions to carbon monoxide gases inside the reactor build­ that must be the improved safety of all ing). Some of the radioactive fission products commercial BWR power plants could eventually be released to the surrounding considered in through work conducted in the environment. BWR Severe Accident Tech­ The goal of the BWRSAT Program is to provide these analyses nology (BWRSAT) Program best-estimate analyses of all of the phenomena in is, perhaps, for the Nuclear Regulatory the scenario described above. Particular emphasis Commission (NRC). This is placed on predicting the timing of key events without work. centered in OR L's (e.g., the time lags between the initial loss of Engineering Technology coolant and events such as reactor vessel and parallel in any Division, is providing new primary containment failures) and one ·rimating insights into the phy ical the environmental releases of fission products. other field of phenomena and processes that The scope of phenomena that must be consid­ like ly would occur in the event ered in these analyses is , perhaps, without parallel simulation." of a severe or ··core-melt" ' in any other field of simulation. These phenomena accident in a commercial include neutronic . thermodynamics. heat transfer. BWR power plant. mass transfer. fluid flow. metallurgy, chemistry. aerosol transport, combustion phenomena, and Accident scenario. One scenari o for such structural mechanics. This is obviously a formi­ accidents a umes the complete failure of a ll dable challenge for both analyst and computer. normal and emergency core-cooling systems, an extremely unlikely event. The failure of these New codes for the Cray. The basic simu­ systems, coupled with the continued boiloff of lation tools employed in the BWRSAT Program water in the reactor, would gradually reduce the are a new generation of computer codes reactor's water inventory and eventually expose (BWRSAR, CO TAIN, and MELCOR), which the hot nuclear fuel. Once uncovered, the fuel incorporate state-of-the-art simulations of the temperature would increase significantly, leading appropriate phenomena, and the Cray X-MP to melting of the reactor core and subsequent supercomputer at the Oak Ridge Gaseous Diffu­ breach of the reactor vessel. sion Plant, which enables engineers to use these After reactor vessel failure. the hot core debris new codes realistically and economically. would fall onto the concrete floor of the primary The BWRSAT work uses these tools in combi­ containment, causing complex chemical reactions nation with detailed plant descriptions (models) to that release a mixture of combustible gases provide analysis capabilities undreamed of just a (hydrogen and carbon monoxide). aerosols. and few years ago. ORNL researchers have pioneered

12 Oak Ridge National Laboratory REVIEW studies of the ability of reactor buildings to retain CONTAIN and MELCOR typically execute 5 to fission products released from the primary I 00 times faster on the Cray X-MP than on the containment during severe accidents. This IBM-3033 system. capability has recently become the focu of much attention within the NRC and the nuclear power Reactor buildings contain releases. In industry. some scenarios, as much as 98% of the aerosols Current ORNL investigations are evaluating the released into intact reactor buildings may be ability of BWR reactor buildings to withstand the retained, never reaching the surrounding environ­ pressures and temperatures generated by the ment. The results of recent analyses indicate, combustion of hydrogen and carbon monoxide in however, that the structural integrity of reactor the reactor building following primary contain­ buildings may be challenged in some accident se­ ment failure. The computer code used for these quences. Much work studies is MELCOR-a 500,000-line FORTRAN remains to be done in code developed by Sandia National Laboratories this field, and many for the NRC. Although MELCOR simulates a uncertainties persist. "Codes such as wide variety of phenomena (including core heatup, However, ORNL's fuel melting and relocation, reactor vessel failure, resource triad of experi­ CONTAIN and core-concrete interactions, primary and secondary enced and skilled per­ containment response, and fission product and sonnel, tate-of-the-art MELCOR aerosol transport), only the containment analysis severe accident simula­ typically capabilities of the code are exercised in the present tion oftware, and the studie . Cray X-MP ensure our execute 5 to A typical ORNL MELCOR secondary contain­ future as a national ment model might consist of 26 computational center for BWR severe 100 times cells or control volumes that represent the various accident research.­ regions within the reactor building. MELCOR Sherre/1 R. Greene, faster on the performs mass and energy balances for each cell; Engineering calculates cell pressures and atmospheric and Technology Division Cray X-MP tructural temperatures; simulates hydrogen and than on the carbon monoxide deflagrations; and tracks the flow of gases, aerosols, and fission products IBM-3033 through the reactor building. Within the MELCOR model, the primary containment blowdown can be system." as umed to enter one or any combination of cells. The accident analysis period of interest for these calculations typically ranges from I to I 0 h, and the computation time depends on factors such as model formulation and the physical phenomena chosen for investigation. Computing time on the Cray generally ranges from 30 min to I 0 h. A recent series of calculations required 6 h of computing time for simulation of a 2-h transient event. Such simulations would not be feasible on less powerful computers. In fact, codes such as

Number Three 1988 13 - SUPERCOMPUTERS IN RESEARCH Supercomputers in Materials Science Theory

has been recognized for over half a century greatly advancing scientist ' understanding of the hat the phy ical and chemical properties of cohesive properties of such materials. materials ultimately depend upon the elec­ There are two important factors underlying these Itronic bonding interactions among the component advances. One is refinement of the local spin atoms, as described by quantum theory. But density approximation (LSDA), a theoretical understanding the cohesive propertie of concept that relates the total energy of an atomic materials remained a chall enge to theorists, unti l system to the electronic charge density. The the breakthroughs in theoretical technique and second is great improvement in computer capacity computer performance achieved during the past and performance. The LSDA provides a reliable decade and a half. As a re ult, the total energy of and tractable framework for describing the interac­ most crystalline solid can now be calculated, tions between electrons during bonding, while recent advances in computer design have provided the means to carry out the complex calculations involved in first-principles studies. r------Using these advanced tools-the LSDA concept and supercomputers-materials scientists have made great progress in understanding cohesion in materials where the atoms occupy ordered sites on an underlying crystal lattice. However, some of the most interesting materials science problems have remained beyond the scope of contemporary theoretical techniques, because most materials contain impurities, defects, and areas of structural disorder. Only in the last few years have the advanced computer programs and computer power been developed to treat such problems. Materials science theorists are now beginning to understand the atom ic-level factors that are important in designing materials with pecific improved mechanical propertie .

Modified nickel aluminide alloy.An out tanding example of an ORNL development in this area was the design of a strong, ductile, new • nickel aluminide alloy for high-temperature service. This superior material was developed, (c) ( d) using microalloying procedures, by the Alloying Behavior and Design Group of ORNL's Metals Contour plots of orbital charge density for strongly bonding and Ceramics Division. The leader of thi group, states in octahedral nickel clusters: (a) bare nickel and clusters C. T. Liu, recently received a 1988 E. 0. Law­ containing (b) lithium , (c) carbon, and (d) fluorine atoms. The plot rence Memorial Award from the U.S. Department plane bisects the octahedron intersecting four nickel atoms and of Energy for his work in alloy development. the impurity atom at the center. Adjacent contours differ by a Before this recent work was done, Ll 2-phase factor of 2, with in itial value= 0.0001 (contour 1). nickel aluminide alloy had very limited applica­ tions, because the materials are intrinsically too

14 Oak Ridge National Laboratory REVIEW brittle to be fabricated into useful structures. The -10 grain boundaries (interfaces between grains in the solid material) are structurally disordered regions that attract impurities from the bulk material, causing embrittlement and severely affecting performance. The key to controlling this grain -14 boundary brittleness in the ORNL nickel / • aluminide alloy was the di scovery that minute / /' additions of boron to the correct alloy composition '/ >Q) resulted in strong interfacial cohesion. The new ->- -18 ORNL-designed material is ductile and can be <.9 easily fabricated. Five companies have already a: w obtained nonexclusive licenses for the commercial z use of the improved aluminide alloy. w <.9 Why boron? At first the question of why z -22 0 boron enhances interfacial cohesion in grain z boundaries of the alloy seemed beyond the scope co of conventional theoretical techniques. However, I c have recently developed, collaboratively with Frank Averill of Judson College, illinois, a new -26 atomic-cluster theoretical approach called the augmented Gaussian orbital method, which solves the LSDA equations for model atomic clusters of arbitrary composition and structure. This approach -30 had the potential for explaining why boron 2.5 3.0 3.5 4.0 4.5 5.0 5.5 produced this remarkable enhancement of cohe­ sion (whereas most impuritie cau e embrittle­ D (bohr) ment). With our new technique, we addressed the Binding energy curves for various problem of explaining the effects of boron and impurity atoms in nickel host clusters. other impurities on the properties of nickel aluminide. First using various atomic clusters to model the grain boundary region of the nickel aluminide alloy, we then calculated the effects of alloys, we first calculated the binding energy as a various impurities on the binding energy and function of interatomic separation for several in teratomic forces in the clusters. bare-host clusters (nickel, aluminum, and mixed metal) as reference systems. The calculations Value of the Cray X-MP. Great care and were repeated for impurity-containing clusters to accuracy are required in all phases of our work, determine the effects of inserting impurity atoms because the energy differences of interest are in in octahedral ho t clusters representing the open the parts-per-million range. Supercomputers such sites where impurities bond in grain boundaries as the Cray X-MP are extremely useful in per­ (see figure above). forming these intensive calculations. To under­ Our research showed that the effects of variou stand how various impurities affect the cohesive impuritie on the cohesive and mechanical strength of grain boundaries in nickel alum in ide properties of the host material are profoundly

Number Three 1988 15 SUPERCOMPUTERS IN MATERIALS SCIENCE THEORY

different. The figure shows the binding energy Superconducting oxides. The tudy of curves we calculated for various impurity atoms superconducting oxides represents another [hydrogen (H), carbon (C), lithium (Li), and materials area where supercomputers will play an aluminum (AI)] when they occupy the interstitial important role. Nancy Wright, Bill Butler, and l have initiated studies in the Metals and Ceramics octahedral ite in a Ni 6 host cluster. The curve labeled Ni 6 Division to understand the interatomic forces represents the bare-host refer­ underlying structural stability of the high-tempera­ ence curve. ture superconducting ceramics. The complexity of Various impurities affect the these ystems will require extensive computing re­ "One key to cohesive properties (indicated sources for performing the calculations involved. our continuing by the curve depths) and One key to our continuing advancement in mechanical properties (as applying electronic structure theory to more advancement defined by the energy variation realistic models for complex materials science with Ni-Ni atom separation) in problems i the availability of high-performance in ... very different ways. Both Li and computers. In a sense, the computer serves a the AI impurity atoms have a de ta­ experimental tool of the theorist, allowing new materials bilizing (embrittling) effect on concepts to be explored through calculation . Our science the nickel host clusters, while H analysis of electronic tructure calculations is and C bind strongly and have a greatly aided by computer visualization of the problems is stabilizing effect on the host. result , such as the electron orbital densitie An analysis of our calcul ation shown in the figure on page 14. Future advance the availabil­ results enabled us to identify the in this field will bring u nearer the ultimate goal atomic-level factors important in of designing materials to have specific de irable ity of high­ determining these effects of im­ qualities, based on an understanding of the atomic­ purities on interfacial cohesion. level interactions involved.-Cay/e S. Painter. performance In particular, the results from Metals and Ceramics Dil'ision computers." our study of boron and su lfur in nickel clu ters provided the first reasonable explanation of why boron enhances cohesion in grain boundaries of superall oy nickel aluminide, whereas mo t impurities, such as sulfur, are embrittling species.

16 Oak Ridge National Laboratory REVIEW SUPERCOMPUTERS IN RESEARCH Fusion Calculations on Cray Supercomputers

ny aspects of fu sion energy From a mathematical point of view, the MHO research require large and model is defined by a system of seven partial complex computations. For differential equations in time and three spatial Mexample, the ORNL studies of the dynamics or coordinates, with boundary conditions given on a motion of fluid plasmas magneticall y confined in toru ·. The intrinsic three-dimensionality of th e three-dimensional configurations (nonlinear problem demands large amounts of computer data magnetohydrodynamic, or MHO, studies) were storage. The problem is further complicated by made possible on ly when supercomputers became the di sparity in the time scales in volved. The available to perform th e necessary computations. instabilities can develop within a micro econd. The MHO model as umes that the plasma is an However. to assess their conse- electricall y conducting fluid, which can be quences fo r plasma confinement, confined inside a magnetic field because it tends we must follow their evolution to flow along field lines but not across them. The for times on the order of I 0 to "Through most common magnetic field configurations I 00 s. This determines the currently used in fu sion research are toroidal maximum time step for the calcu­ the use of (doughnut-shaped) fields that hold the pl asma lati on and implies that the within a finite region, away from the fusion vessel number of time steps in the com­ advanced walls. putation is on the order of lOx. The figure on page 18 shows a numerical and Plasma instability problems. The torus cross secti on at several sudden growth of instabilities in a plasma is a times in the calculated evolution computational major problem in keeping it confined long enough of a large-scale instability. Before techniques and for net fusion energy to be produced. Con idering the in stability is triggered, all the plasma as a flu id, we can imagine that vortices magnetic field lines li e on nested the Cray-2 develop; like wire loops moving in a magnetic tori (shown as slightly shifted fie ld, these vortices induce currents in the fluid. In circles). As the instability grows, super­ tum, these localized currents create perturbations the magnetic topology changes in the confining magnetic field. The current loops and large magnetic " islands'' computer, we in a magnetic field also exert forces on the fluid de velop (from field lines that can now solve that may either reinforce the initial vortex (the wander radially in an organi zed unstable case) or damp it (stable case). In the fashion). Because of the nonlin­ problems of unstable case, the magnetic field perturbations can ear interactions between the be large enough to change the direction of the fie ld large- cale modes of in stability, MUD lines, causing them to wander radi all y. They may chains of smaller islands are even touch the vessel wall, resulting in a los of generated (t = 1300). When turbulence plasma confinemenL the magnetic islands overl ap ORNL researchers are investigating these (t = 1358 to 1398), the magnetic that were plasma instabilities and their consequences for line wander radially in a chaotic previously plasma confinement in an effort to find ways to fashion and confinement is lost in avoid them. These investigations will increase our most of the plasma volume. impossible." understanding of the confinement properties of pl asmas in ex isting fusion devices such as The California Crays. ORNL's Advanced Toroidal Facility and guide the Because complex studie uch as design of more-advanced experimental devices. these are not possibl e without supercomputers.

Number Three 1988 17 FUSION CALCULATIONS ON CRA Y SUPERCOMPUTERS

the creation of the National Magnetic Fusion Energy Computer Center at Lawrence Livermore National Laboratory (LLNL) was critical to the development of this fie ld. The Center provide two Cray-1 computers and a four-processor Cray- 2 for fusion research and a two-processor Cray X­ 0 MP for energy research. ORNL's Fusion Energy Division normally uses the Cray-2 for the nonlin­ ear MHD calculations. This supercomputer ha a memory of 64 million words, which is shared by its four processors. For 1988, the Department of Energy ha allocated 1800 h of Cray-2 computing time to ORNL's Fusion Energy Division. The complexity of the calculations requires us to optimize our u e of even thi s large allocation of time. Multitasking our calculations, or dividing them into simpler tasks that can be performed independently and in parallel on the four processors, is an important step in this optimization that leads to reduced costs for memory and throughput time. Personnel in the Martin Marietta Energy Systems Computing and Telecommunications Division have worked closely with Fusion Energy Divi ion theorists to develop a multidisciplinary approach to this nonlinear MHD research. Through the use of advanced numerical and com­ putational techniques and the Cray-2 supercom­ puter, we can now solve problems of MHD turbu­ lence that were previously impossible. For thi s 0 work, we solve the equivalent of a half-million coupled nonlinear differential equations for 10,000 independent time steps. This level is still short of the computational demands expected for parame­ ters of new fusion experiments with hotter

-1L-~-L~~~~~_L~--~ plasmas. However, the information gained can -1 0 1 -1 0 be usefully compared with experimental measure­ X ments at the plasma edge. The r~sults will Torus cross sections calculated for various times in substantially enhance our current understanding the evolution of a fusion plasma instability. of the basic turbulence mechanisms and will make valuable contributions to the development of analytical fusion theory.-Vickie Lynch, Energy Systems Computing and Telecommunications Division, and Ben A. Carreras, Fusion Energy Division

18 Oak Ridge National Laboratory REVIEW SUPERCOMPUTERS IN RESEARCH Theoretical Atomic and Nuclear Physics

he advent of modern heavy-ion accel­ complex ity of the problem and results in a tors and supercomputers has revolu­ si mpler, flowing-fluid description of the ni zed atomic and nuclear physics. coll ision: wide variety of new phenomena have already been discovered using these advanced We interpolate the fields, or fluid elements, tools. Scientists in ORNL's Physics Division can 0 usi ng a linear combination of basis splines­ now stud y colli sions between nuclei as massive as mathematical devices introduced in the last uranium at relative speeds very near the speed of century as a means for describing the light and distances ranging from about the size of contours of smooth objects such as the atoms to a scale much smal ler than the individual hulls of boats; protons and neutrons that make up an atomic nucleus. 0 We solve the mean field problem by evaluat­ In studying the most energetic nuclear collisions, ing the fields on a special space lattice. A an interesting new feature has emerged: for short diagram of the points on such a lattice is periods of time during the col li sion, the strengths reminiscent of a crystalline solid. of the electromagnetic fo rces (which hold the atom To illustrate, the figure on this page represents a together) and the nuclear forces (which hold the near-collision between uranium nuclei (solid nucleus together) become nearly equal. One of the circles) in a three-dimensional space lattice whose new structures produced during this type of coordinate axes are shown in the lower left. The collision is a vacuum consisting of strongly shaded region between the nuclei depicts the coupled particle-antiparti cle pairs. The new induced magnetic field. vacuum state decays, usually in < I Q- 19s, with the emission of th e particle pair . Experimentalists Lattice field equations. In principle, the detecting the emission of these pairs are alerted to fu ll quantum-mechanical, many-body equations the fact that something unusual has occurred to of motion should yield an exact description of the disturb the norm al vacuum. To interpret and verify what the experimentali st measures, it is essential to have a reliable means for computing the same phenomena using various theories. The strong electric fields produced in these collisions can be as large as 1025 V/m, exceedi ng any other laboratory-produced electric field and comparable to the fields in the center of a neutron star. In contrast, the largest electrostatic fields that have been produced by conventi onal mean were on ly -107 Y/m, in ORNL's Holifield Heavy Ion Research Facility. The enormous nuclear-collision fields preclude the use of traditional analytical approaches for understanding the novel effects y that appear. We found it necessary to completely reformulate the problem-solving method, to exploit the special advantages of supercomputer architectures. Our analysis method involv es three steps: z Owe first consider an approximation. based on Representation of a colliding atomic or a mean fie ld picture, whi ch reduces the nuclear system in a space lattice.

Number Three 1988 19 SUPERCOMPUTING IN RESEARCH

"Our results nuclear collision, but the of the vector architecture of OR L' s Cray X-MP lead us to equations are not currently supercomputer and the matrix processing of the solvable. However, a simple Floating Point array processor. These are the two believe that model can provide us with a principal supercomputer used by ORNL ·s remarkably accurate view of the Physics Division, and both machines can olve super­ process. We treat the collision the lattice field equations of our studies with between the two heavy ions or high efficiency. computers nuclei as a fluid dynamicist The floating point FPS 164/Max array processor, have an treats the movement of complex wh ich is acces ible from the Physics Division bodies through a viscous fluid. VAX, is a pipeline machine with independent immense Each nucleon or electron, as a addition and multiplication processor units. The fir t approximation, i consid­ present configuration has 2.5 million word of fast potential for ered to move independently in memory and two matrix coprocessors, or Max the field of the other particles. boards. Each Max board con ists of two addition nuclear and As each particle changes its and two multiplication processors. The basic position, the field it create con­ machine cycle time yields addition or multiplica­ atomic tinuou ly changes. This is tion speed · of about 5.5 MFLOPS (M illion physics usually called a mean field FLOating Point instruction per Second) for each picture of the collision, and the processor so that, as presently consti tuted, the applications." resulting motion of the nucleons system has a maximum cycle capacity of 55 or electrons is much like that of MFLOPS. The specialized matrix multiplication is a classic magnetic field or fluid. handled quite differently on the Max array It is necessary to solve the processor sy tern than on a vector pipeline mean field equations of motion machine uch as the Cray. without any other approximation. The wavefunc­ For example, a test calculation was carried out tion of nucleon and electrons, and hence the on both the Cray X-MP and the FPS 164/Max, mean field , must be determined at each space ach ieving 70 MFLOPS on the Cray and 40 point, .r.y.:, and at each time, 1, as the collision MFLOPS on the Max. Although the Cray required proceeds. The equations governing the motion of only half as much time to complete our test the mean fields are complicated nonlinear calculation, we found the use of the Max array differential equation . which will not be di cussed processor to be more cost effective because in thi article, but we can imagine the colliding charges for Cray use at ORNL are greater than system placed in a box, as illustrated in the figure those for use of the array processor. In fact, we on page 19. Space inside the box is denoted by a could not have completed this work without the mesh of points .r,. yi" :,. The size of this box availability of the low-cost supercomputing depend on the problem under con ideration: it capacity of the FPS 164/Max. Our results lead us would have to be tens of fermis (I o-1.• em) acros to believe that supercomputer have an immense at the distance of clo e t approach of the colliding potential for nuclear and atomic physics applica­ nuclei , and 10' times larger than this to describe tions, because so many computations in these the initial atomic states. The equations of motion areas can be reduced to such simple matrix governing the mean field can be repre ented by operations a we have described.--Christopher lattice field , or matrix, equations that can be Bottcher and Michael R. Strayer, Physics solved in a straightforward manner with a variety Dirision of supercomputers.

Vector and matrix coprocessors. To solve our matrix equations, we have developed algorithms that exploit the respective advantages

20 Oak Ridge National Laboratory REVIEW SUPERCOMPUTERS IN RESEARCH Parallel Computers for Efficient Problem Solving

omputer users continually demand Not all types of problems lend themselves more computing power. Even on the to a parallel solution. However. our experience current supercomputers, some rel a­ shows that a number of computati onall y expen­ tivelyC small problems can be intractable. For sive problems can be efficiently solved on example , to accurately simulate the heat transfe r in parallel computers. For example. ORNL's Intel a l-m .1 block using a simple computation method hypercube, a di tributed-memory multiprocessor, req uires 4.8 x 10 11 arithmetic operations per and the Sequent Balance 8000, a shared-memory second of time simul ated . For a computer to do the multiprocessor. have been used for the numerical simulation in real time would require a sustained approximation of problems in molecular dynam­ computation rate I 000 times greater than that ic s, stress analysis, magnetohydrody nam ics, ac hievable by a Cray X-MP. contaminant transport, and electrostatics. All of To meet the need for greater process ing peed, these problems can be posed as partial differential computer designs have emerged that make equations (PDEs), and parallel computers seem to multiple processors avai lable to the computer user. be particularly well-suited to their numerical and almost all mainframe computers incorporate approximation. In our work we have developed some sort of parallel process ing. Existing parallel methods fo r so lving several common type s of computers are generally one of two types: shared­ problems, performing the exhaustive calculations memory multiprocessors and di stributed-memory on a di stributed -memory multiprocessor. multiprocessors. The hared-memory type allows all processors equal access to memory, usually Parallel solution of PD Es. Usually the through a common communication channel, or solutions of partial differential equations are ·'bus.'' As the number of processors sharing a bus functions over in creases, the ma xi mum di stance between a some spatial processor and the memory increases, and so does domain-for the potential for contention between processors for example. the control of the bus. Both conditions degrade the velocity of a fluid communication speed. Thus bus architectures in a pipe or the cannot efficiently employ a large number of electromagnetic processors, although commercially available potential on a computers of this type exist with up to 30 plate. The solution processors. may al so vary Distributed-memory multiprocessors are charac­ with time. When teri zed by a network of communication channels, solving such each of which connects a processor to other equations. it is processors and memory units. Typically each sufficient to know processor has a local memory of limited size. the solution at When a processor needs data stored in a memory some finite time other than its own , the computer program must point and loca- explicitly request transmiss ion of those data. Extra tions within the programming effort is required for coordinated spatial domain. Example gri d in unit square. data movement. Because communication channels For example, if are not shared in di stributed-memory architec- the soluti on is defined on the unit sq uare, then it ture , they are more effic ient than bus architec­ may be enough to know th e value of th e solution tures in the use of processors and memory. But the at the points of intersection of the square with lack of equal access to memory by all processors some uniform grid (see figure ). makes the use of distributed-memory computers Partial differential eq uat ions express a relation­ more difficult and more limited. ship between the solution function and the

Number Three 1988 21 PARALLEL COMPUTERS FOR EFFICIENT PROBLEM SOLVING

problem data. A standard by u ing averaged data at approach to numerically 0 0 0 0 time 0 from five nearby approximating the • locations in the grid, if r is solution at the grid not too large (see the locations is to 0 • • • 0 adjacent figure). replace the POE (Xj, Yk) For this sample prob- with an alge­ lem, the approximation of "A clever braic equation 0 0 0 0 the solu tion at each that relates the • location in the grid assignment solution values requires the same amount at the grid of computation, and it is of tasks to points to the problem data. This alge­ best to compute the solution in parallel by braic relation is a discreti:ation of the L:J partitioning the grid into equal-sized quare processors POE. If orne care is taken, the olution block of locations and according of the discrete problem approximates the corresponding solution of the partial U letting each processor calculate the olution to location differential equation. It is the di cretized values associated with one block. POE that must be solved using parall el Data for locations in the interi or of a block are can mini­ computation. needed by onl y one processor, while data for The efficiency of this computation is locations on th e boundary are required by more mize the influenced by two factors: the than one processor. Thus, dividing the computa­ computer's "load balance" and commu­ tion according to this partition of the problem time spent nication costs. If work is distributed domai n limits the sharing of data by the processors unevenly, then most of the processor forwarding and make the parallel computation efficient. may be idle during much of the compu­ For a distributed-memory architecture, addi­ tation time- a waste of the computer's messages to tional information is required. To specify the power. Even if the work is well ­ algorithm, we must other distributed, processors will be idle while they are waiting to receive the data they D assign the blocks of solution val ues that are processors need to perform the next task. They are near each other in the spatial domain to also effectively idle when they are too processors near each other in the computer as well as busy sending and receiving data to start and the next computation task. These the time assign the data associated with a particular communication costs are strongly L:J block of locations to a memory location near influenced by the rel ative position of spent wait­ the proce or that is calculating the as ociated the processor within the network . A oluti on values. ing for data clever assignment of tasks to processors according to location can minimize the Because most of the data are needed in only one to arrive." time spent forwarding messages to other processor, they should be located as near as processors as well as the time spent possible to that processor. The data that must be waiting for data to arrive. shared with other processors will require less computer ti me spent in transmi sion if all the Explicit methods. Solving a POE by an processors needing this information are near its explicit method involves computing the ol ution memory location. The optimal implementati on of value at a given location directly from data for any explicit algorithm wi ll depend both on the nearby locations. For example, in a heat-transfer architecture of the multiprocessor being used and analysis, the temperature for time r , at location on the domain of the problem. (Xj, Yk) in a grid, can be accurately approxi mated

22 Oak Ridge National Laboratory REVIEW Implicit methods. Unfortunately, explicit Conclusions. Neither the standard direct algorithms are not always the best means to ap­ nor the standard iterative methods for solving proxi mate the solution of the PDE. Generally, the PDEs were designed to be efficiently executed on approximation describes a relationship between a multiprocessors. But our experiences using th e solution value, the nearby data values, and the ORNL parallel comput- nearby solution values. The discretization of the ers indicate that these PDE is then represented by a system of coupled multiprocessors are linear equations, also called a matrix equation. applicable to a wide Thus the value of the solution at a given point is range of energy-related "Our only implicitly defined for each' time step. To research problems, and calculate the solution values, we must use the staff members of the experiences parallel algorithms for solving simultaneous linear Mathematical Sciences equations. This is currently an area of intense Section of EPMD are using the mathematical research and, fo r the most part, currently investigating beyond the scope of this article. We will describe new algorithms designed ORNL only some general approaches. to exploit processing Techniques for solving matrix equations in parallelism. Problems parallel parallel differ depending on whether the matrix requiring the solution of computers has many nonzero entries (a "full " matrix), or only PDEs seem particularly a few (a "sparse" matrix). The research of Mike suited to the efficient indicate that Heath , AI Geist, and Chuck Romine of ORNL' s utilization of parallel Engineering Physics and Mathematics Divi ion processors, and we have these multi- (EPMD) has led to excellent methods for the only briefly outlined solution of matrix equations when the matrix is some methods for processors full, but the matrices for most PDEs are sparse, utilizing parallelism in are having perhaps only 10% nonzero entries. solving them. As the There are two major approaches to solving commercially available applicable to sparse-matrix equations. The first, which calcu­ parallel computers lates an exact solution, is simi lar to the algorithms increase in both number a wide range for full matrices and is referred to as a direct and speed of processors, method. Alan George and Esmond Ng, also of the they will increasingly be of energy­ EPMD, have been developing direct sparse-matrix applied to the solution of algorithms for parallel computers. computationally related The second major approach finds a solution to intensive problems.­ research sparse-matri x equations by itera ti v~ly improving John B. Drake and an estimate of the solution. Iterative methods are Patrick H. Worley, problems" attractive for parallel processing, because each Mathematical Sciences step of the iteration is relatively si mple and can Section, Engineering often be efficiently implemented on a multiproces­ Physics and Mathemat- sor. A particularly attractive method of this type, ics Division known as the preconditioned conjugate gradient method, has been implemented by a number of researchers on the ORNL hypercube. While iterative methods can work very well in parallel, they are not best for all problems. Typically, three­ dimensional fluid dynamics problems are solved using iterative methods, while two-dimensional solid mechanics problems are solved usi ng direct methods.

Number Three 1988 23 • Rising aspirations and popu lation growth in developing countries such as Pakistan (shown here) add up to growing needs for energy services. The U.S. Agency for International Development, assisted by ORNL, is trying to help meet these needs. Energy for Development: ORNL Returns to the Third World By Thomas J. Wilbanks and Sherry B. Wright

he 1980s have been a quiet decade Since 1982. Oak Ri dge ational Laboratory has r energy research and deve lopment been involved in analys is and tec hnical as istance & D). as th e energy crises of the 1970s to help deve loping co untries mee t th ei r energy ave faded into th e background . But needs, as a part of the programs of the U. S. lower prices for oil in world markets have not Agency for Intern ational Development (AID). In solved th e energy probl ems of deve loping A sia, A frica, Latin A meri ca, and th e Middle East, co untries . where a ·hortage of reliabl e and ORNL staff members from seve ral divisions are affordable energy services has continued to low working with their counterpart from th e countries th e engine of soc ial and economic deve lopment. involved and with oth er co lleagues to help solve At the same time. th e growing appetites of th ese current energy problems. countries for commercial fuels have made th em Of course , helping th e deve loping world is not a th e fas test growing market in th e world for energy new ve nture at OR L. From 1959 through 1965, technologie. , and many observers believe th at th e Oak Ridge School of Reac tor Technology their energy dec isions will have a profound effect trained 11 2 individuals from 27 countrie - on the world ·s total energy requirements (and th e including Braz il. India. Indonesia, Pakistan, the global environment) by the middle of th e next Philippines . and Thai land-in nuclear reactor century.

24 Oak Ridge National Laboratory REVIEW ORNL subcontractor Jean-Roger Mercier and Energy Division 's Robin Cantor record the weight, cost, and quality of a typical Malagasy family's charcoal purchase to meet a day's cooking needs. This kind of information is hazards evaluation and operations superv ision. In resources were needed at home. and ORNL 's being tracked addition, OR L staff member John Pinajian, connections with developing countries decreased. longitudinally by Fritz McDuffie, Frank PlasiL and others lived and In the 1980s, however, the situation changed as AID and the World worked in developing countries as representatives energy was sh ifted to the back burner as a domes­ Bank to reveal of the former U.S. Atomic Energy Commi sion. tic policy issue. ORNL needed innovative ways to trends in actual During this same period, ORNL was becoming keep its energy expertise fresh in order to be household fuel a leader in examining the applications of ''big prepared to meet domestic needs certain to lie costs throughout sc ience," such as the "Nuplex" concept. to ahead. One response was to look for energy needs Antananarivo, de velopment problems at a grand scale. As in other countries linked to U. S. government Madagascar's conce iv ed by ORNL, a Nuplex would supply programs and the U. S. national interest. capital and inexpensive power from a large generating AID, the U.S. government agency responsible largest city. facility-probably but not necessarily a nuclear for as isting developing countries , had developed plant-located on an arid eacoast in a developing significant energy programs during the 1970s. In region. A desalting faci lity would make fresh 1982 a number of discussions among AID, the water from the seawater, pump this water inland to U.S . Department of Energy, and ORNL resulted irri gate the desert, and provide electricity and in an interagency agreement wi th DOE to have thermal energy for the communiti es and agro­ ORNL conduct a one-year energy planning and processing industries that would follow ( ee article assessment project in Liberia for AID. in the Review, Summer 1967, pages 3-9). In the process, ORNL attracted world attention as a Energy Planning for Liberia center of desalination expertise, stemming from the Laboratory's Molten Salt Reactor Program. By Liberia is a smal l developing country on the the early 1970s, ORNL was well-known in Asia west coast of Africa, about I 0% smaller than the and Latin America. and Laboratory leader Murray state of Tennessee, with a population of approxi­ Rosenthal and others vis ited developing countries, mately two million people. In 198 1, the gross such as Pakistan, to provide technical advice. domestic product (GOP) of the country was Because of the mushrooming demand for $84 1 million for the monetary economy and energy R&D and problem solving in the United about $1 billion for the monetary and traditional States during the 1970s, the Laboratory's economies combined (or roughly $500 per

Number Three 1988 25 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

capita). This measure of development ranked director of what was then the Liberian Bureau of Liberia a little below Bolivia, Zambia, and Hydrocarbons (now the Department of Energy) Indonesia but a little above Pakistan and the within the Ministry of Lands, Mines, and Energy. ,, Sudan. During the late 1970s, the country' long­ Seven ORNL staff members participated in the as- standing favorable trade balance had disappeared, essments that year. "The most as the bill for oil imports grew from $5.5 million By normal standards for such projects, the striking in 1971 to $152 million in 1980 (in nominal impact of that year's work was dramatic. Beside dollars). The country was mired in a severe producing a mass of fre h data. whi ch has been ORNL con­ recession; real GDP was falling at the rate of adopted by Liberian in stitutions and international tribution . .. 5% per year, and national indebtedness wa lending institutions alike as a standard base for showed that rising rapidly. evaluating energy alternatives. the project changed It wa natural to look to the energy sector for the direction of energy policymaking in Liberia. Liberia some of the solutions to Liberia's growing The most striking ORNL contribution was a study, should shut economic problem , but attempts in the late 1970s initially sketched out by Dick Barne and later to identify initiatives in thi s sector had foundered amplifi ed by Garland Samuels and other Energy down its oil because of a lack of reli able information about the Division taff member , that showed that the refinery country's energy sy tern. its need , and its country should shut down its oil refinery rather options. At the reque t of the Liberian govern­ than borrow money to upgrade it. rather than ment. AID agreed to support a project to improve An inefficient, simple, poorly maintained borrow the base of information and to strengthen fac ility, the refinery wa producing gasoline, money to Liberian capabiliti e for energy planning. In turn, diesel fuel, and other light petroleum products at a AID a ked OR L to conduct the project, in very high co t compared with world market prices. upgrade it." collaboration with Liberian coll eagues and ORNL analyses showed that if, instead of refining consu ltants from two universities and a private oil within the country, Liberia imported oil firm. products at world prices and sold them domesti­ The initial project. a national energy assess­ cally at the hi gher prices to which domestic ment, was carried out over a one-year period in markets had become adjusted, the country would 1982 and 1983. At the time the project began, a realize a revenue windfall of $15-20 million per hi ghl y factionalized energy policy-making system year. This was equivalent to about a 2% increase had led to a focus on three proposed actions: in the GDP-a substantial payoff, indeed, from a (I) large-scale development of hydroelectric $200,000 project! After this assessment wa power, whi ch centered on the St. Paul Project, a confirmed through an inder,endent study by an proposed storage and generation facili ty (with an international consulting firm, the refinery was shut ultimate capacity of -1000 MWe) at the conflu­ down. As a result, Barron reported, the Liberian ence of two rivers upcountry; (2) renovating and government was able to tri ple the tax on gasoline upgrading Liberia' only oil refinery to increase and diesel fue l while reducing prices at the pump its efficiency and hift its product slate toward by 10%. li ghter fractions (at an e timated co t of $50 Other contributions of the project may be million); and (3) domestic oil explorati on and equally important in the long run, even if they are production, given the country's proximity to o il less dramatic in the short run. For example, the production offshore from the Ivory Coast. project team showed that the propo al for the large Working with Liberia's ational Energy Com­ St. Paul hydroelectric project was based on a mittee (NEC), the U.S.-Liberian project team demand forecast that ignored uncertainties about produced a summary assessment, 18 working the country's rate of economic development papers, and a number of informal comments and during the next 20 years. A more appropriate adv isories during that first year. The team wa led strategy would be to add generating capacity in by Bill Barron of ORNL ·s Energy Division and smaller increments, allowing it to be fine-tuned to M.-H. Neufville, secretary of the NEC and the development actually experienced (e.g., first

26 Oak Ridge National Laboratory REVIEW upgrading an existing hydroelectric facility). fueled facilities for power supply in rural areas is As another example, the project team drew being actively explored, and the NEC is conduct­ attention to Liberia's impre sive wood resource, ing energy planning and analysis on a continuing which could meet a sizeable share of the country's basis without external assistance. energy needs on a su tainable basis. e pecially in rural areas. Particularly interesting is the prospect ORNL's Role Expands "ORNL staff of using wood gasification to power stationary conducted internal combustion engines, which account for By 1984, AID's Office of Energy (see figure two-thirds of Liberia's oil consumption. on page 29) was finding the entire program of short-term Other examples include the introduction, during which the Liberia project was a part-the energy analyses an electric power shortage in December 1982, of a planning and policy development (EPPD) portion number of ideas about energy conservation and of a larger program entitled Energy Policy of still­ load management and the tran formation of the Development and Conservation--difficult to unresolved NEC, which had previously been an ineffective manage. Without a staff arrangement for techni­ issues, such body, into a major force for rational energy cal assistance, the program lac ked coherence and planning. was not producing the desired results. as the At the end of the first year, the government of AID decided that ORNL, which had performed economics of Liberia and AID decided that further support was impressively in Liberia, might be able to solve needed to help turn the project's information, this problem. The Laboratory was world-famous harvesting ideas, and recommendations into a national as a diver ified center of energy R&D. including the wood energy trategy. Barron moved to Liberia for two energy policy analy is, and during the late 1970s, resource year to serve a technical advisor to the NEC. it had developed a strong reputation for program ORNL staff provided support by conducting short­ management (assisting busy federal program and the term analyses of still-unresolved issues, such as offices in connecting with the best available potential for the economics of harvesting the wood re ource technical resources in the larger re ·earch and the potential for energy conservation in community, based on the Laboratory's own energy buildings. This period culminated with the technical strengths). With these qualities in mind, conservation i suance in June 1985 of An Integrated National AID asked the Laboratory to lead and manage the Energy Program for Liberia, submitted by the EPPD project and, with the agreement of DOE, zn NEC to the government of Liberia and adopted as ORNL accepted this new challenge. buildings." the nation's energy strategy. In September of that It seemed to ORNL that the fir ·t step should be year, just before Barron's return, he organized a to take a fresh look at energy conditions in retreat for high-level official of the Liberian developing countrie in the 1980s. For example, government, energy institutions in that country. by 1984, three years after EPPD program major lenders, and others to di scuss and publicize directions had been defined. oil prices had leveled the strategy. and the dominant view within AID was that "the Soon afterward, a controversial election was energy problem .. had gone away. at least as a held in Liberia and most development assistance major priority for allocating scarce development from AID, the World Bank, and other lenders was resources. In addition. a major AID-sponsored placed on hold while the political situation was conference in Reston, Virginia, in February and sorted out; thus, much of the trategy is yet to be March 1983, held to take stock of the experience realized. But it remains the road map being used with national energy planning activities in by all parties interested in Liberia's energy future developing countries, had concluded that, in and the project itself was judged by AID to be one many cases, national energy planning wa not of the most succes ful on record. The refinery helping to meet energy needs. Quite simpl y, remains closed, the St. Paul project has been although the planning was sometimes being done replaced by a successful rehabilitation of existing reasonably well. it was seldom connected with hydroelectric facilities, the possibility of wood- actual decisions about major energy actions and investments.

Number Three 1988 27 ENERG Y FOR DEVEL OPMENT: ORNL RETURNS TO THE THIRD WORLD

ORNL's Work for AID Is Global in Scope LATIN AMERICAN ASIA AFRICA MIDDLE EAST CARIBBEAN

LIBERIA HAITI PAKISTAN ZAIRE COSTA RICA PROJECT INDIA TUNISIA MADAGASCAR ECUADOR ACTIVITIES BANGLADESH JORDAN SUDAN GUATEMALA PHILIPPINES MOROCCO* TO DATE MALAWI CARIBBEAN* INDONESIA* KENYA PERU*

WORKSHOP/ SRI LANKA NEPAL BRAZIL BOTSWANA CONFERENCE THAILAND CHILE PARTICIPATION MALAYSIA

PROJECT ACTIVITIES INDIA UPCOMING

PROJECT NIGER PAKISTAN BOLIVIA ACTIVITIES SOMALIA EGYPT PHILIPPINES HONDURAS UNDER SUDAN DISCUSSION

*Through subcontractors

Out of a series of white papers by ORNL and Another emerging problem that has ignificant several subcontractors. which were prepared as a ramifications for AID program planning wa also part of the Office of Energy's first comprehensive identified. Increasingly, the development commu­ program plann ing exercise, a djfferent view took nity is recognizing that the best development paths shape about the energy problem in developing are sustainable ones. In other words, a strategy for countries in the 1980s and of the chall enges to development that exploit option that look AID and other development as istance agencies attractive in the short run , but cannot be sustained in their efforts to help with solutions. The studies over a longer term, can teer a country toward targeted fou r major energy problems in most painful adjustments and changes, often just as countries: capital requ irements for power system economic growth is starting to take off. Clearly , ex pan ion, needs for energy price reform, needs such a trend should be anticipated and avoided if for adequate and affordable hou ehold fuels, and possible. In the energy ector, thi concern has at inadequate energy ystems to support rural least three manifestations: ( I) the continuing development. They also pointed out three general reliance on imported oil; (2) wood fue l use problems in the process by wh ich energy needs of associated with deforestation and ecological developing countries are addressed: insufficient damage: and (3) the growing use of coal, espe­ attention to private-sector roles, a lack of atten­ cially in A ia, when management of the global tion to the proce s of implementing good ideas, environment may call for some constraints on and a failure to learn enough from experience. fossil fuel emissions.

28 Oak Ridge National Laboratory REVIEW ------INTERNATIONAL DEVELOPMENT -- COOPERATION AGENCY

/ ~ OPIC TOP

Asia/Near East Africa Latin America/ Bureau Bureau Caribbean Bureau \ Field Missions '-. Other '-. '-. directorates '-. '-.

Meanwhile. the 1986-1988 program plan of the Sciences. Engineering Technology, Health and Office of Energy ca ll ed for an expansion of EPPD Safety Research. Fusion Energy. and Computing activities related explic itl y to the new prioriti es; and Te lecommunications divisions and the former Organizational AID field missions were showing considerable Information Resources Organization. along with chart for AID. interest in accessing ORNL 'spool of technical ind ividuals from more than 90 subcontracted resources through AID's Office of Energy. In firms. universities. other national laboratories. addi ti on . in 1985. the Laboratory was also asked and consulting practices. Project activities have to assume responsibility for a second Office of taken team members to 26 developing countries. Energy program concerned with renewable energy wit h ac tivities upcoming in another and under applications and training (REAT). This combina­ discuss ion for six more (see table on page 28). tion of developments has resu lted in substantial This rather sizeable and diverse activity is growth in ORNL-Ied AID projects. ORNL directed by Tom Wilbanks ofORNL's Energy activities have reached $2.5 million per year in Division. with substantial help from AI Ekkebus, funding (see figure on page 30). with some of the Larry Hill. Ed Hillsman. Don Jones. Bob Perlack. recent growth reflecting support from AID fie ld Carl Petrich. John Reed. Bob Shelton. Dan missions and regional bureaus. So far. the AID­ Waddle. and Sherry Wright. Current directions supported work has involved 60 ORNL and and emphases. flowing directly from the analyses Martin Marietta Energy Systems. Inc .. staff and planning conducted in 1984-1985, include members from the Energy. Env ironmental power for development. household fuels, and in st itutional and policy reform .

Number Three 1988 29 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

planning. For instance, in many cases the cheapest 3 .0 r------~ way to provide more c::::::::J Other Subcontractors/Consultants electricity service (e.g., El} comfort, convenience, ~ ORAU r::: mobility, productivity 0 EL'Z)In-House .E increa es) is not to build 2 .0 another power plant but co to improve the efficiency r::: .E of the power system: 0 existing power plant .s operations, transmi ·sion en ....(].) and distribution net­ .a 1. 0 works, and end uses. '6 r::: During the late 1970 and (].) a. early 1980s. many U.S. X electric utilities became UJ significant innovators in increasing efficiency. It seems certain that power 1982 1983 1984 1985 1986 1987 1988 1989 (projected) system planners in Fiscal Year developing countries can benefit from thi experi­ ence. Power system efficiencies in developing countrie are usually quite low (see figure on page 31 ), and adding new Power for Development generating capacity-the approach assumed by the The AID program Throughout the developing world, electricity World Bank estimates-usually entails unneces­ at ORNL has consumption is growing; annual growth rates in sarily high capital costs for the services added. For grown rapidly. the range of I 0 to 12% are common. According to example, a rudy conducted by Ed Hillsman of the World Bank, this trend translate into a ORNL, David Jhirad of AID, and Pirooz Sharafi requirement for $500 to $900 billion in capital of Hagler, Bailly, and Company. Inc., indicates investment in the next I 0 years for power sy tern that an estimated capital requirement for power expansion in developing countries, or else the y tern expansion in India of more than $50 billion development process wi ll be inhibited-a cri is in the next five year could be cut to $30 billion or indeed. If a olution is to be found for this power less by successful implementation of a least-cost crisi , it lies in some combination of three approach to planning. strategies: ( I ) rationalizing energy prices to Putting such an approach to work in developing reduce demands resu lti ng from inappropriately countries does not just depend on convincing the low prices, (2) increasing private sector roles in countries themselves of its merits, however. There power sector financing, and (3) improving the is considerable evidence that the "least-co t" efficiency of systems th at provide electricity options are not alway the most easily financeable services to minimize the cost of increa ing options, and developing country decision makers the e services. are more likely to respond to signals from lenders ORNL's power system planning program than from analysts. Accordingly. the AID-ORNL focuses on the third option: improving system program has al o opened up a dialogue with the efficiency through " least-cost" power ystem major development lender , to pass on these perspectives.

30 Oak Ridge National Laboratory REVIEW Work at ORNL on the issues regarding power for ~ 20% development is also losses concerned with two other energy problem : power for rural areas and §§§ sustainable paths for - 71% energy system develop­ generation ment. The current empha­ losses sis of activities related to 1 rural power systems is on fuel burned technological and institu­ to generate plant and line c::> end-use tional options for mini ­ inefficiencies efficiencies grids (i.e .. medium-scale elecrcity ):1 alternatives to extending the central grid, on the one high(2x) - 7 % hand, or elf-generation, losses - on the other). Bob Perlack ~ of ORNL has helped to develop a trategy to §§§ - pur ue uch options as: persuading a re latively generation losses large private-sector rural enterprise, for example a mineral s extraction or agroprocessing plant, to generate enough power to meet needs of other users and se ll it to a coopera­ Paki stan National Coal Conference in 1986. at tive or other third party for distribution in a which Shelton, Wilbanks. and Eissenberg made • Power system minigrid. Perlack and others have also analyzed presentations. R. Krishnan and others from the effi ciencies in the pros and cons of substituting a simple mini­ Engineering Technology Division are ass isti ng an developing grid, connected to a small centralized power R&D in stitution in India in improving capabilities countries are generator. for small decentralized equipment for for evaluating experiments with fluidized-bed usually qu ite low, water pumping. Advantages of the mini grid option coal combu. tion. ORNL is leading AID's effort placing heavy include power for small rural industries (especiall y to determine the potential of coal as a wood-fuel demands on when requirements for water pumping are small ) substitute in developing countries. Add itionall y, energy and capital and for other village needs. as well as simpler ORNL is joining with Bechte l and Pyropower resources. For operation and maintenance. Finally, David to assess the feasibi li ty of using oil shale to example, an Eissenberg of OR L 's Engineering Technology generate electric power in Jordan. efficient system Division has played a major role in preparing a Second, the Laboratory-in association with (bottom) delivers 2.5 times more basic guidebook on technologies for water lifting Oak Ridge Assoc iated Universities' Institute for electricity se rvices in developing countries. Energy Analysis-has taken stock of more than a from a given ORNL 's program for AID includes three kinds dozen years· ex perience in renewable energy amount of fuel of activities related to sustainable development. applications in developing countries to focus AID than an inefficient First, the Laboratory is helping to explore the role programs more effectively on projects and system (top). of coal and other solid fossil fuels in energy technologies that have high probabilities of strategies for developing countries. ORNL success. Under the leadership of Dan Waddle, organized the technical program for the first ORNL is he lping to initiate new projects in several countries.

Number Three 1988 31 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

ORNL and ORAU­ IEA have assessed experiences with a variety of renewable energy technologies in Third World countries over the past 12 years. Here a photovoltaic system (designed by the National Aeronautics and Space Administration and Lewis Research Center) provides a West African village with power to supply drinking water, lighting, and grain milling. (Photo courtesy of Third, ORNL has helped to develop a new wi th a coal resource. Seventeen AID-assisted VITA.) program component for EPPD that is intended to countries fit th is description, and coal briquetting increase attention to environmental management is a fam ili ar, well -establi shed technology (usuall y in energy project planning in developing in vo lvi ng pyrolysi , except where the source coal countries. is anthracite). In 1985, ORNL developed a generic approach for evaluating the potential of individual Household Fuels for countries for coal-briquette-substitution. This Wood-Scarce Areas approach was first applied in Haiti, the mo t defo rested country in the We tern hemisphere The basic options fo r better meeting the (and the poorest). The prototype feasibility study, household energy needs of people in wood-scarce led by Glenn Stevenson, uncovered a market in the areas of developing countries, besides encourag­ Port-au-Prince area fo r the output of a 50,000- ing more tree cultivation, are ( I ) finding alterna­ tonne/year coal briquetting plant. But it also tives to wood as ources of energy, (2) fi nding concluded that briquette made from the best­ alternative source of wood energy, and known deposit of Haitian lignite would not (3) improving the efficiency with which wood is compete successfull y with wood charcoal at produced and used a an energy source. Since current charcoal prices. Briquettes made from coal 1984. OR L in vestigations have focused on the of higher energy value. however, whether from potenti al of ·ubstituting clean-burning coal import or other deposits in Haiti, mi ght be briquette fo r wood charcoal in wood- carce areas competitive even with inexpensi ve charcoal.

32 Oak Ridge National Laboratory REVIEW 1. 9 400

1.8 380 ... ORN L's Energy Cl) Division analyzed Cii 1. 7 360 a! 0 a; the costs and () E :u 0 benefits to a C) 0 Q) :0 .r::; 1: :J 1. 6 340 ..... typical rural vi ll age irrigation load .....0 ~ Q. ., .!!! in Senegal if it E c: 0 were to meet its ::I Q) 1. 5 320 ~ ll. ~ needs for irrigation water pumping through a 1.4 300 centrali zed power generati on system 1 . 3 '------'----'---'------'----'---'------'----' 280 coupled to a 0 5 10 15 20 25 30 35 40 "minigrid" rather Levelized load factor than continuing to (percent) use small independent plus agroprocessing diesel pump sets. w If village amenities >en p lu s school/c linic and small local -c 0~ --g industries devel op ~< c o plus cottage industries ~0 o - to use off-peak Z.J :.Jc: power from a w :0 .Q plus village lighting (/).J ~(6 central facility, w< c .g> plus cold storage overall pum ping a:: a: :.=... .: a.:::J costs decline until wa:: plus households the unit cost of a: power is lower plus rural industries than the current level.

Discussions of further tests of this idea are under hybrids from coal and residues. Moreover. 1985 way with private-sector entrepreneurs. assessments of household fuel options in Zaire ORNL also led a recently completed effort to and Madagascar, conducted in collaboration with evaluate markets for coal briquettes in Pakistan. the World Bank, prompted a more detailed ex­ The assessment indicated that, with some care in amination of the possibility of producing charcoal selec ting high-energy source coals and with high­ from alternative sources of wood-that is, from density urban markets, coal briquetting is an source areas or species that could be used without attractive Pakistani business proposition. A exacerbating deforestation. In early 1986, briquette production operation, however, would ORNL's Carl Petrich and Robin Cantor, together need to penetrate firewood and/or kerosene with three consultants, assessed the potential of markets to some degree (in addition to the converting smallwood from a large pine planta­ relatively small charcoal markets) to justify tion in Madagascar to charcoal to slow down de­ bui lding a 50.000-tonne/year plant. forestation in the capital city area. The re search Increasingly, ORNL's household fue ls program effort included a pioneering (in developing has been expanding to cover other options as well, countries) use of " focus groups" to evaluate the including briquettes from agricultural residues and market acceptance of pinewood charcoal. The

Number Three 1988 33 A nearby windmill pumps drinking water to this central storage tank, which supplies a village in Mali (photo courtesy of VITA) .

34 Oak Ridge National Laboratory REVIEW ------

team concluded that. although pinewood carboni­ good ideas are major impediments to meeting zation for the general urban charcoal market faced energy needs, institutional and policy reforms are a number of obstacles, the potential was signifi­ important in most developing countries. In this cant. The World Bank and the government of connection, the Laboratory 's research and techni­ Madagascar are currently negotiating a major loan cal assistance for AID are focused on energy to go forward with such a project. price reform, problems in the implementation of In 1988, ORNL has been evaluating options good ideas. and technology transfer. for improving the efficiency of household fue l Most of the professional attention to energy production and use to reduce wood requirements. price reform in developing countries to date has A remarkably successful effort to produce and sell been devoted to estimating the appropriate price. more-efficient cookstoves in Kenya is being ex­ general ly related to the marginal cost of produc­ amined to identify lessons that can be transferred ing the energy. Frequently, however, thi process to other countries, and markets for improved cook­ has actually been counterproductive, because the stoves in urban areas in Central America are also estimated ·'rational" price is so much higher than being assessed. the current price; consequently, price reform often seems to local decision makers to be Institutional and Policy Reform po litically and economicall y risky. ORNL's work has been directed instead toward developing Because energy pricing problems. insufficient approaches to energy price reform that can be private-sector ro les, and problems implementing implemented realisticall y. A recent stud y by

Local private entrepreneurs in Kenya have found that manufacturing and selling more energy-efficient cookstoves can be a good business opportunity .

Number Three 1988 35 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

Larry H ill. for example, proposed a generic they are identified. Jn fact, literally hundred of Charcoal is the strategy that combines gradual price increase good ideas have been proposed in recen t years cooking fue l for with (e.g .. energy fac ility investments offering pay back most fam ilies in period of 18 month or less) but not acted on, improvements in the efficiency of energy Haiti, and charcoal U while le appropriate actions have been taken. supply institutions (to red uce the marginal consumption is The problem seems especially acu te where th e cost of producing energy and distribu te the associated with right choices involve innovation, because new bu rdens of price reform to suppliers as well severe options are difficult to eval uate and, if they as consumers), deforestation . involve financial and institutional changes, they ORNL has are harder to put into action. Clearly, ORNL's assessed the U improvements in the efficiency of energy u e activities related to power for development and potential for (so that the price of energy services will ri se manufacturing and le s rapidly than th e price of energy), an d energy price reform are focused on this problem. marketing a clean­ In addition, ORN L ha been contributing to the burn ing coal U change in in titutional tru ctures assoc iated developmen t of a major new AID project in India, briquette in Haiti with energy pricing (so that more appropriate the Project to Accelerate the Commercial ization of as a substitute fo r price leve ls will be se lf-sus taining). Energy Research (PACER). A five-year effort wood charcoal. budgeted at more than $20 million, th e goal of thi (Photo courtesy of As this indicate , a general problem in develop­ project is to develop be tter connections between Mike Benge, ing countries is implementing good ideas once indigenous Indian energy R& D institutions and USAID .)

36 Oak Ridge National Laboratory REVIEW Buses are impo rtant publ ic transportati on in most developing count ries. Anal yses by ORNL and others have shown that oil fuel consumption can be reduced by transportation system efficiency improvements.

the country's electric power system. The hope Other Activities is that such institution-building will speed the process of communicating and evaluating new As long as this list may seem to be, it still technology options for improving electricity neglects a host of other recent ORNL-AID services. ORNL will serve as an important source activities. For example, the work of David Greene of U.S. technical assistance. In addition, ORNL is (Energy Division) and others on the potential for as isting AID 's Office of Energy in taking a more improving the energy efficiency of transportation comprehensive look at the implementation systems in developing countries, based on field problem. investigations in Tunisia and Costa Rica, is Beside· representing a significant attempt to considered the best available reference work on break new ground with implementation, the this subject. In one major contribution, Greene, PACER project is also an experiment in technol­ ORNL 's Frank Southworth, and colleagues from ogy transfer. In other project activities, ORNL is Hagler, Bailly, and Company, Inc., conducted encouraging technology transfer by seek ing experiments in Costa Rica that established for the private-sector participation in coal briquette first time that driver training and improving production in Haiti and Pakistan, improving the maintenance practices can pay off in significant state-of-the-art for market acceptance studies in reductions in fuel use by bus and taxi fleets (see developing countries, assisting AID with informa­ table on page 38). tion dissemination , and providing insights about the special challenges faced by U.S. energy technology producers in the competition for markets in developing countries.

Number Three 1988 37 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

%FUEL SAVINGS BY CONSERVATION OPTION

DRIVER DRIVER INCENTIVES PROPER RADIAL TOTAL** TRAINING (BUSES) MAINTENANCE TIRES

BUS FLEET 4.2* 1.2 6.5* -1.8 11.5%

TAXI FLEET 15.1* 1.8 0.2 16.8%

*Statistically significant results **Totals include other conservation options that are not listed here

In other AID activities, ORNL L:J conducts policy analy e for AID. U exploring the potential of coal and natural gas as transition fue ls from oi l to renewable Driver training and L:J assists with energy project design and energy sources. which depends both on tech­ improved eval uation (i ncluding recent project evalu­ nology R&D and environmental assessment: maintenance of ation efforts in Bangladesh, the Philippines, and bus and taxi fleets Ecuador, and the Sudan), in developing Uidentifying and helping to fill technology countries can yield L:J conducts an energy policy research program gaps for the longer term (e.g., the clear need significant fuel for AID's Office of Energy (whi ch issues for better, cheaper, cleaner technologies to savings, according requests for proposals and award contracts meet relatively large energy requirements in to analyses by for small research projects to provide infor­ developing countries in the next 25 to 50 ORNL and Hagler, mation and maintain a research ba e for AID years ). Bailly , and programs), and Company, Inc. We also see opportunities for the Laboratory to L:J provides a variety of technical information contribute in other fields of science and technol­ and decision support assistance to AID, ogy. The best prospects seem to be in environ­ rangin g from technical information assis­ mental management, a rapidly growing area of tance from AI Ekkebus and the Admini tra­ concern in the developing world, and in uses of tive Services organization to technical assis­ microcomputers and modem decision support tance from Sherry Wright and others with systems in developing countries (e.g .. electric office automation and decision support utilities could use microcomputers to reduce systems, including the development of new delays and inefficiencies in billing). Other possi­ decision support tools for the Office of bilities include water resource issue , especially in Energy. Africa, and transportation system management, especially in the increasingly clogged urban areas Outlook of developing countries. Our hope is that, in addition to helping ORNL Given all the uncertainties, forecasting the stay in the energy business during the 1980s, this future of any energy program i risky. But the kind of direction for the Laboratory will add to our prospects for further ORNL contributions to impact in solving real problems of real people. For meeting energy and other needs in developing most of the ORNL staff members who have been a countries are bright. As we look ahead in our part of the program so far, the greatest satisfaction program development, we see particular opportu­ has come from knowing that they could directly nities in and personally make a difference in improving the L:J power system development--connecting lives of others. power system planning with technology More generally, of course, the world faces a advances ranging from generation to end-use, half-century in wh ich two great transformations

38 Oak Ridge National Laboratory REVIEW Thi s mule-d ri ven kerosene "tanker," tethered to a power pole in Pakistan, illustrates the energy changes taking place in developing countries.

must occur at the same time-an economic trans­ science and technology, long-lasti ng rather than formati on in developing coun tries and a transfor­ transitory, and of undeni abl e importance to mation in energy systems away from such a heavy society. Because of the Laboratory's in terests re li ance on petroleum and natural gas. Achieving and activ ities in developing countries si nce th e both of these transfo rm ati ons concurrently is one late 1950s. and especiall y si nce 1982, ORNL of th e great chall enges to sc ience and technology has the opportunity to play a major ro le in in our time. This challenge is ti ed directl y to the meeting thi s challenge. we ll-being of a majori ty of the world population and, thus, to such larger issues as world peace. (The authors would like to acknowledge the ORNL has traditi onall y looked for chall enges such research assistance for this article provided by as this-big. ri ch, complex, multidisciplinary. Steve Olive, now affiliated with the University foc used on energy and the environment. rooted in of Hawaii.) =:1

Number Three 1988 39 ENERGY FOR DEVELOPMENT: ORNL RETURNS TO THE THIRD WORLD

Biographical Sketches

Thomas J. Wilbanks (center) is an ORNL Corporate Fellow and senior planner for ORNL's Energy Division. He hold a Ph .D. degree in economic geography from Syracuse University, is the author or coauthor of five books or monographs and more than 50 journal articles and professional papers, and has served on a variety of national and international panels and committee related to energy policy and technology assessments.

Sherry B. Wright is a technical associate in the Energy Divi ion and serve as the general manager of ORNL project work for the Agency for International Development (A ID). She holds a B.S. degree in business education from Cumberland College.

Carl Petrich, who provided substanti al assistance to the authors in assembling illustrations for this article , is a research staff member in the Energy Divi ion , a key member of the AID project team, and a speciali t in graphic de ign and communication. He has an M.S. degree in re ource planning and landscape architecture from the University of Michigan.

40 Oak Ridge National Laboratory REVIEW Take A Number A Prisoner's Puzzle

A prisoner is given an opportunity to participate in a random-selection experiment, the outcome of which may set him free. H e is given two id entica l urns and 10 green ball s and 10 red balls. He is asked to distribute the 20 balls into both urns. When the prisoner has fi nished, one of the urns will be chosen at random, and the prisoner will be asked to pick a ball at random from this urn. If the ball is green, he w ill be set free. The question is: How can the prisoner best apportion the balls in the two urns? It turns out that the prisoner's best strategy is to put one green ball in one urn and the re­ maining 19 balls in the other urn. With that distribution, the probability that he will be set free after the experiment is nearly 75 %.

lOOth Power of Natural Numbers Take the natural number 7, raise it to the lOO th power, and note that it ends in 1. Similarly, raise 2 to the l OO th power and note that it ends in 6. It can be shown that when any natural number is raised to the lOOth power, the results end with either digits 0, 1, 5, or 6. If 5 is raised to the l OOth power, the final two digits are 25. In fact, when any natural number is raised to the lOOth power, the last two digits can be only 00, 01, 25, or 76. Even more interesting, the final three digits of the l OO th power of any natural number will be either 000, 001, 625, or 376.

Number Three 1988 4 1 Designer Steels for Ron Klueh (left), Ph il Maziasz, and Bob Swindeman discuss the behavior of designer steels after creep testing Energy in the Multi- Frame Skutt Creep Testing Facility behind them . Applications Creep behavior of small specimens By Carolyn Krause (like the one held by Swindeman) is used to predict the lifetime behavior iet revolution in the design of new U A modified 12% chromium, 20% manganese of machine rainless steels and other iron-based steel that. compared with conventional steels, components (l ike all oys for use in advanced, high­ is expected to be both resistant to radiation the tubes in the Atemperature energy systems is under way at Oak damage and to have a much greater and more foreground) in Ridge National Laboratory. Since 1984, ORNL rapid decay of radioactivity induced by long­ energy production researchers in the Metal and Ceramics (M&C) term exposure to intense neutron irradiation in facilities. Division have designed several outstanding new a fusion reactor. Fusion reactor components alloys: made of this "low-activation" material would pose less of a radioactive disposal problem at U A modified 14% chromium, 16% nickel the end of their useful lives. stainless steel, whi ch has excell ent resistance to radiation damage-swelling and severe The developers of these alloys are metallurgists embrittlement caused by prolonged irradia­ Phi l Maziasz, Ron Klueh , and Bob Swindeman of tion at elevated temperatures-and creep­ the M&C Division. Together they have expertise the long-term tendency to stretch at high in advanced electron microscopic analysis along temperatures under an appli ed stress. This with an understanding of the mechanical behavior material, which has up to I 000 times the and fabrication processing of iron-based steels creep resistance of type 316 and other com­ and alloys. mercial stainless steels at 700°C, could be Their alloy design is based on knowledge used in both advanced coal power plants and gained from intensive years spent stud ying creep nuclear reactors. and other effects of neutron irradiation and thermal exposure on alloys. ln particular, they U A modified all oy that consists of 20% have made use of new in formation about irradiated chromium, 30% nickel, and less than 50% alloys gained from analytical electron microscopy iron by weight (hence. the designation alloy (AEM), which can provide structural and rather than steel). Resistant to corrosion as compositional information at the 2- to I 0-nm well as creep, this alloy may prove to be scale. The new alloys are a direct spin-off of particula.-:y u eful for superheater and ORNL alloys already designed and developed to reheater boiler tubes in advanced steam improve resi tance to radiation-induced ·void cycles for coal-fired power generating swelling and helium embrittlement in fusion systems and other hi gh-temperature uses. energy applications.

42 Oak Ridge National Laboratory REVIEW Maziasz calls these new materials "de signer deliberately and carefully controlled to preserve stee ls" or '"gourmet all oys" because their or add good properties. Decisions about what and properties are predetermined by adding unique how much to add are based on the special insights combinations of key trace elements and because gained from studies of the effects of long-term knowledge of material behavior is essential for radiation on steels using AEM. This information deriving the right alloy "recipe." Understanding has suggested ways to design alloys to meet the complexities of all oy behavior during use is special needs by altering their microstructures." the key to designing steels. Just as a gourmet cook (See sidebar on page 48.) inventing a new dish must know the criti cal ingredients and understand their possible interac­ tions during cooking, the metallurgist designing a Gourmet Alloys new alloy must know the likely contribution of In 1976, Jim Stiegler, now director of the M&C each ingredient in specific proportion to the alloy Division, started a group to demonstrate the recipe during use as well as the synergistic effects concept that a material's microstructure is linked of combined ingredients. Will interactions with to its properties, suggesting that manipulation of other constituents improve the all oy enough to the microstructure could achieve desired proper­ achieve a desired characteristic? Or will they ties. In the past 12 years, several groups in the degrade the alloy's desirable properties? If a little division have shown that these ideas are correct. is good, would a lot more be better--or worse? In 1983, ORNL 's Magnetic Fusion Energy ORNL metallurgists have shown that minor Materials Program, headed by Everett Bloom of changes in the composition of alloys can change the M&C Division , encouraged Maziasz to try to their microstructure enough to produce desired design a steel for fusion reactors that has better properties, just as adding a touch of thyme and a radiation resistance than the titan ium-modified pinch of garlic can improve an ordinary tomato steels that had been developed at ORNL in sauce enough to please a discriminating gourmet. 1977-1978. Design engineers want to build "Small changes make a big difference," Maziasz fusion reactor "first walls" from a steel that resists says. "For years, M&C researchers have added small amounts of titanium to steel to produce fine metal carbide precipitates that improve radiation or creep resistance. In recent years, Arthur Rowc li ffe and Eal Lee have observed that add ing small amounts of phosphorus to steel Maziasz peers at improves its radiation resistance. an irradiated steel specimen in a Adding small amounts of boron has Philips CM30 been known for years to improve analytical electron creep res istance in steels and has microscope, one been found at ORNL to improve the of th e key ductility of nickel aluminides. instruments for "However, the secret of our understanding success has been the realization that how alloys of combinations of these elements are specific more effective than any single compositions addition by itse lf. The al loys devel­ change during irradiation . oped at ORNL are not produced by a mysterious black art or random guesswork. Minor additions are

Number Three 1988 43 Ed Bolling performs cyclic fatigue testing of an advanced austenitic steel specimen heated by inductive eddy currents.

damage by the 14-MeY neutrons produced by very good swelling resistance, as measured by a fusion reactions of hydrogen isotopes in plasmas. new remote precision densitometer designed Maziasz' recipe for a modified stainless steel specifically for such measurement on small , has proved doubly uccessful: the new al loy highly radioactive specimens by Lloyd Turner and exhibited both better radiation resistance and Roy Buhl. excellent high-temperature thermal creep-rupture What is the recipe for this steel? Says Maziasz, strength. Radiation resistance is resistance to "If you had I 00 grams of this steel, it would swelling and embrittlement--cau ed by point contain 64.4 grams of iron (Fe), 14 grams of chro­ defects produced by displaced atoms and by mium (Cr), I 6 grams of nickel (Ni), 2.5 grams of helium produced by transmutation--during molybdenum (Mo), 2 grams of manganese (M n), intense, prolonged neutron irradiation. Creep­ 0.5 gram of vanadium (V), 0.05 gram of phospho­ rupture strength involves a combination of rus (P), 0.01 gram of boron (B), 0.3 gram of (I) resistance to strain (stretching) accumulation titanium (Ti), 0.1 gram of niobium (Nb), and 0.1 during long, high-temperature exposures when gram of carbon (C)." In technical horthand, the stresses below the tensile yield stre (often quite all oy are called new Fe-14Cr-16Ni-2.5Mo-2Mn small ) are applied and (2) resistance to embrittle­ steels modified with carefull y controlled additions men! and premature fracture. of V, P, B, Ti, Nb, and C. In 1984, Tom Roche of the M&C Division This new modified 14Cr-16Ni steel could also made the first mall heats for reactor irradiation be u ed for fast-breeder reactor core and structure experiments and limited tensile testing. Irradia­ applications. In addition, high-temperature tensile tion of mall disk in ORNL's High Flux I otope te ting (done by Noble Rouse of the M&C Reactor for the Fusion Materials Program showed Division) and microstructural analysis of unirradi-

44 Oak Ridge National Laboratory REVIEW Maziasz shows one of the weights used to "stretch" a steel specimen inside the machi ne durin g a creep test. Hanging weights at left indicate that a creep test is in progress.

ated specimens of these same experimental steels to produce electricity more efficiently. An showed good strength and precipitate structures, advanced steam-cycle fossil power plant has suggesting that these steels might also resist not been built in the United States since the straining during longer-term creep testing and at Eddystone Unit I went into operation for the lower stress levels. This behavior inspired Philadelphia Power and Electric Company in thoughts of a spin-off of these tee! from fu ion 1960, because materials then avai lable could not and breeder application to some new high­ withstand the high temperatures and tre es of temperature, nonnuclear application. long-term operation. On June 15, 1984, Ron Bradley, then manager From mid-1982 to mid-1984, the Fossil Energy of ORNL's Fossil Energy Materials Program, Materials Program assessed the materials needs of attended an M&C Division staff meeting during advanced steam cycles. To build a highly which Maziasz presented a talk on advanced efficient, reliable steam-cycle plant, the ORNL austenitic steels for irradiation resistance in steam­ researchers concluded, it is critical that the super­ cycle materials. He suspected that the radiation­ heater and reheater tubes in the boiler be made of resistant steel that Maziasz had developed was a material that can with tand the damaging effects also creep resistant. Bradley expressed interest in of fire (on one side) and steam at 650°C (on the this material becau e of its potential use in other). Plant operation and lifetime depend on advanced steam-cycle, power generating systems various criteria, one of which is the ability of that burn pulverized coal. A number of countries. these materials to resist excessive creep or including the United State , are interested in cracking. Alloys such as CR30A (trademark of building such systems. which, compared with Nippon-Kokan Steel Corporation, Japan) and conventional fossil power plants, use hotter steam alloy 617 could be used for this purpose, but they

Number Three 1988 45 DESIGNER STEELS FOR ADVANCED ENERGY APPLICATIONS

are ex pen ive because of their high chromium creep resistance on the basis of its designed and nickel content (alloy 617 also contain s cobalt microstructure. and molybdenum, which, along with chromium, In September 1987. at the International Confer­ are strategically important for defense and ence on Advances in Material Technology for industry and subject to supply interruption · Fossil Power Plants in Chicago, Maziasz and becau e of political instabilities). Consequently, Swindeman reported that their modified 14Cr- DOE funded an effort to develop an a ll oy that is 16Ni stainl e steel had twice the rupture trength of type 316 stainl es steel, 50% more strength than alloy 800H, and a measurable increase in strength over that of the reference 17 -14CuMo Microstructures of alloy. The creep mill-annealed strengths of these ORNL materials, creep teels could meet that tested at 700°C required for an advanced and 170 MPa. The superheater alloy at photos on the right 700°C and approach the compare the superalloy creep trength microstructure of of alloy 617. While a conventional alloy stainless steel such as 800H with that of type 3 16 can withstand the new, modified, stresses of about 40 MPa 800H alloy (20Cr-30Ni) at 700°C for 100,000 h developed at without rupture, ORNL's ORNL. The lean steel can withstand marked up to I 00 MPa at 700°C differences in for the same period. precipitate '"lean'' (lower in chromium, nickel, cobalt, and However, it was anticipated that the new lean microstructures molybdenum), less dependent on foreign steel would be suitable fo r advanced steam-cycle are directly suppli es, les expensive than superalloys, and plant only if it could be protected by claddings or responsible for better in its high-temperature creep resistance coatings to improve its corrosion resistance. So the property than the reference copper-molybdenum ( 17- Swindeman and Maziasz decided to pursue a differences. 14CuMo) all oy used in Eddystone Unit I . parallel effort to modify corrosion-resistant all oy In the summer of 1984, the Fossil Energy 800H to improve its creep resistance, which is Materials Program and M&C Divi ion staff only marginally better than that of type 316 developed a propo al for alloy development for stainless steel. Maziasz and Swindeman readjusted superheaters for advanced steam cycles. Thu , the proportions of minor alloying element addi­ Maziasz began collaborating with Swindeman tions while changing the Cr-Ni content of 14Cr- and Gene Goodwin of the M&C Material 16Ni stainless steels to that of alloy 800H, which Joining Group for Welding Studies to test his ha 20% Cr and 30% Ni to provide corrosion all oy to determine if it coul d have fos sil energy resistance. The original all oy 800H has a composi­ applications. Because only small heats of material tion somewhat between that of a superall oy had been made at OR L for reactor testi ng, (titanium and alumin um are added to provide larger heats were procured from commercial trength) and that of a carbide-strengthened steel. vendors-Combustion Engineering in Chat­ They also readjusted the base composition (by tanooga in 1986 and AMAX in Ann Arbor, adding molybdenum and eliminating alumi num) to Michigan, in 1987-for the Fos il Energy Materi­ make the alloy act more like a steel having fi ne als Program studie . In the first creep te ts run carbides and phosphides for strength . Ba ed on by Bob Swindeman and Ed Bolling, the new findings from hi welding studies, Goodwin sug­ steel exhibited excellent high-temperature gested that phosphorus addition be limited to

46 Oak Ridge National Laboratory REVIEW improve the alloy's weldability. The result was a alloy. with minor all oying modifications to 20Cr-30Ni-2.5Mo-2Mn alloy, modified with produce fine precipitates. is a low-activation steel various minor element additions, several heats of that has good strength and should have good which were also produced commercially by radiation resistance. Calculations of radioactivity AMAX in 1986. induced by fusion reactor expo ure indicate that The new modified 20Cr-30Ni alloy is 100 times this new steel wou ld be 1000 times less radioac­ more rupture-resistant than alloy 800H at 700°C tive 50 years after exposure to intense neutron and 170 MPa and is comparable to the modified radiation than a typical conventional steel (e.g .. lean stainl ess steels. In addition. this corro ion­ type 316 stain less steel). Besides it apparent resistant alloy is also tougher and more resistant to appli cability to future fus ion components. the cracking along grain boundaries during creep than new Fe-Cr-Mn steel also has good strength at alloy 800H. In short, this new designer alloy is high-temperature strength and may soon find well suited for either advanced steam-cycle or other nonnuclear applications. fludized-bed combustion fossil power plants Maziasz says that these alloy successes show because it is highly resistant to both creep and the value of interdisciplinary programs at national corrosion. laboratories. "Development of these specialized From the standpoint of DOE's fusion materials alloys resulted from the combination of our program, Maziasz's 14Cr-16Ni steel scored high expertise and the needs of ORNL ·s fusion and mark s for radiation resistance. Unfortunately, a fossil programs. This cross-fertili zation , made changing set of design and safety requirements made this alloy un uitable for commer­ cial fusion devices because of possible disposal problems. It contains elements uch as nickel , niobium, and molyb­ denum that, when subjected to Solution-annealed prolonged neutron irradiation. steels aged for are transmuted to hi.ghly ra­ 166 h at 800°C. dioactive isotopes having long Note the large half-lives. The material could differences in be too radioactive to be easily precipitate and afely buried once its microstructures useful life has expired. among the various Applying the principles for alloys . ORNL's a priori design of alloys, new steels have a unique combi­ Maziasz and Klueh attempted nation of phases to de ign a low-activation that are more steel that is as strong and finely distributed resi tant to radiation damage as the modified easy by ORNL ·s organizational structure. than precipitates 14Cr- 16Ni-2.5Mo-2Mn stainless steel. The initial produced rapid breakthroughs in the design of in the other alloys. overall strategy for designing low-ac tivation steels novel materials." had been deve loped by Klueh and Bloom several Maziasz, Klueh, and Swindeman are achieving years before. their goal of designing dramatically improved To reduce neutron-induced radioactivity, Klueh steels and alloys for advanced energy applica­ and Maziasz eliminated nickel , niobium, and tions. These alloy designers have shown that molybdenum from their steel and increased the correctly .. pieing up" a material's microstructure manganese content. Manganese is a good substi­ can produce gourmet alloys having the properties tute for nickel because it. too, stabilizes the to meet special need . =:1 austenitic alloy and because its neutron-induced radioactivity decays more rapidly than that of nickel. The resulting Fe-12Cr-20Mn-0.25C base

Number Three 1988 47 A Matter of Precipitates

"If the right composition is selected, precipitates formed in the alloy during use will beneficially alter the microstructure."

uring prolonged exposure of metals rated-that forms particles of new crystal phases and alloys to ne utron radi ati on. from the original homogeneous parent materi al. atoms are di spl aced, creating vacant These precipitate phase particles. whose structure atom sites and inter titial defects (dis­ and compositi on di ffer from those of the all oy Dpl aced atoms in the inters.ti ces between normal matrix, may ti mulate, inhi bit , or bl ock oth er latti ce atoms). These point defects can agglomer­ changes in th e microstructure, thus affecting ate into voids (holes) and dislocati ons th at cause properti es. the materi al to swell and become brittle. Preci pi tate effects are directl y re ponsible for 'The best way to top this process, whi ch is improvements in mechani cal properti es such as activated by prolonged rad iati on, hi gh tempe ra­ creep resistance. If the right compositi on is tures, and applied stresses, is to pin the dis loca­ selected, precipitates fo rmed in the all oy during ti ons with precipitates-th at is, to induce the use will beneficiall y alter the mic rostructure , formati on of parti cles that trap these extended producing a "tailored precipitate microstructure.'' defe cts in th e crystal structure and keep them What is a microstructure? A metal or alloy has from moving throughout the materi aL'' says Phil grains of near-perfect crystal in various ori enta­ Maziasz of the M&C Div ision. ''But, it is easier tions. The submicroscopic structure includes said than done. Precipitati on in irradiated materi­ grains (crystallites) of diffe rent sizes, grain als is hard to study and characteri ze. harde r to boundaries (interfaces between grain s). precipi ­ understand. and even harder to contro l. " tates. and dislocati ons. These mi crostructural Precipitation is an elevated-temperature elements can change considerabl y under the reaction between all oying elements-initia ll y in influence of elevated temperature and/or exposure soli d solution, which then becomes supersatu- to radi ati on.

Fractography by scanning electron microscopy and creep-crack growth rate data from alloy specimens tested at 700°C. The new steels developed at --~--- I' 14 · 16 Mo 55 ORNL (AX-5) are very resistant to t7 14 Cu Mo I {hi A~ ~1 cracking. The 17-14CuMo alloy cracks 1oo•c ,' roo•c more easily because it fractures in a I brittle manner along grain boundaries.

K ( MPo rrTil

48 Oak Ridge National Laboratory REVI EW Laves precipitation along grain boundaries in the 17-14CuMo alloy causes brittle failure along grain boundaries, which shortens the creep-rupture life at 700°C (see upper photo­ graph). By contrast, carbide precipitation along grain boundaries in the new ORNL steels contributes to a dramatically different, ductile failure mode that lengthens the rupture life (see lower photograph).

importance of fine tuning the alloy com­ posi ti on to achieve formation of the desired precipitate phases while eliminat­ ing the unwanted ones." Maziasz cites phosphoru s as an ORNL metallurgists have spent many years example. ''If you add too much phosphorus, the identifying the precipitates formed in steels that steel is hard to weld. But if you add a little phos­ have been irradiated for a long time in reactors and phorus and supplement it with a li ttle vanadium, understanding their effects on steel properties. In it can greatly strengthen the steel.'' 1965, Arthur Rowcl iffe, now in the M&C Division During th e lifetime of a steel, bad precipitate but then a doctoral student at Manchester Uni ver­ phases can form. These can be brittle interrnetal­ sity in England, discovered the existence of lic phases that are detrimental to desired proper­ chrome phosphide preci pitates in steel and ties; for example, "Laves phases" can cause suspected that they coul d improve the creep severe grain-boundary embrittlement. The trick is resistance of austeniti c stainless steels. lronicall y, to alter th e steel composition so that irradiation or almost 20 years later Maziasz, working at ORNL creep aging will produce other precipitates, which on his doctoral thesis and on improving the then prevent formation of these bad phases. For preci pitate stability of titanium-modified austeni tic exampl e, additi ons of carbon and boron stabilize stai nl ess steels, un covered evidence, togeth er with the formation of carbides and retard th e formation Rowcli ffe, that new iron-titanium phosphides also of the harmful intermetallic phases in irradiated play an im portant role in the creep resistance of steels. steels modified with titanium and phosphorus. These findings have been made possible by a Subsequent transmi ssion electron m icroscopy highly sophisticated AEM faci li ty, which all ows work by Maziasz and others showed that certain use of such techniques as scanning transmi ssion fine precipitate phases, such as FeTiP. are much electron microscopy. electron energy-loss spec­ more effective at pinning dislocations than other troscopy. convergent-beam electron diffraction, phases. and the most useful of all for development of " My AEM studies of irradiated and un irradiated these designer steels, X-ray energy-dispersive steels," says Maziasz, ''also verified that some spectroscopy for compositional m icroanalysis. precipitates are rich in certain elements that cause "Our AEM st udies."' Maziasz says, ··provide a desi red effects and block undesired effects in the fountainhead of new ideas about how to alter th e surrounding microstructure. I also learned the microstructure to obtain desired properties." =::1

Number Three 1988 49 New DNA Stain Aids Cell Studies By Luci Bell

uring the past century, scientists uranium or osmium, are routinely used to prepare have learned much about the general (i.e., nonspecific) contrasting reagents for JD) TEM work, but reagents with high specificity are mysterious processes and micro­ scopic structures found inside living cells. generally either unavailable or extremely difficult Although most cells are naturally somewhat to prepare and use. "A recently homogeneous and nearly transparent in appear­ ance, biologists have been able to tudy the A Collaborative Effort completed minute, but complex, internal cell mechanisms by interdiscipli­ developing structure-specific dyes that will cling In a recently completed interdisciplinary project nary project to and highlight the particular cell components at ORNL, chemists worked in collaboration with they wish to investigate. In the mid-19th century, biologists to enhance their means of unraveling the atORNL researchers first began using several natural and details of fundamental cell structures and pro­ allowed synthetic dyes, or staining agents, to prepare cell cesses. Using "seed money" from ORNL's Ex­ samples for microscopic examinations that ploratory Studies Program, the scientists were able chemists to revealed the intricate basic structures of cells. to develop and test a new DNA-specific staining work in Since that time, the knowledge and instrumenta­ agent that significantly improves the use of collabora­ tion used in biological studie have become much electron microscopy as a technique for under­ more sophisticated, but the cytochemical staining standing fundamental cell processes and the tion with agents and techniques have not kept pace with complexities of cellular genetic materials. biologists to this development. The collaboration began when Ada L. Olins and In this century, transmission electron micros­ Donald E. Olins, researchers in ORNL's Biology enhance copy (TEM) has become an important inve tiga­ Division and professors in the University of their means tive tool for distinguishing among fundamental Tennessee-Oak Ridge Graduate School of cell structures such as chromo omes, composed Biomedical Sciences, needed a reliable DNA­ of unraveling of deoxyribonucleic acid (DNA)-the material specific stain for their cytochemical studies. The the details of that reproduces cell characteristics generation Olinses are using a new method of electron fundamental after generation. TEM produces images of cell microscope tomography, which they recently components ba ed on their electron density developed in collaboration with Henri A. Levy cell differences and, since most living cells are of (now retired) of ORNL's Chemistry Division and structures relatively uniform density, the "stain" used for computer specialists in Martin Marietta Energy TEM imaging must adhere to some cell structures Sy terns' Computing and Telecommunications and in preference to others (i.e., be structure-specific) Division. Their major goal is to determine the processes." and inc rease the density of the targeted structures three-dimensional path of DNA in active and enough to make them appear darker than other quie cent chromosomal structures. An effective, parts of the cell (i.e., produce significant con­ reliable, DNA-specific staining agent was trast). Compounds of heavy metals, uch as considered essential for this work.

50 Oak Ridge National Laboratory REVIEW to begin preparing, purifying. and Ada Olins and characterizing Bruce Moyer use these compounds. an electron David Allison. microscope to electron microsco­ view cell pist in the Biology structures stained Division, agreed to by a new DNA­ test the compounds specific agent. for actual staining performance.

A Superior Stain In one of those occasional seren­ dipitous occur­ rences in scientifi c The on ly known DNA-specific stain was an investigations. one of Kim's first preparations unstable. hard-to-produce. osmium-ammine proved superior to Gautier's stain in several complex (a chemical of uncertain composition. but ways: which has been determined to contain both osmium atoms and ammonia molecules in some 0 Sample preparation is greatly simplified, comb in at ion). This osmium complex. called the requiring only a few hours. rather than the Gautier stain. was first prepared and studied. by the usual 4 to I 0 days; Swiss biologist. A. Gautier. The Olinses' request 0 Staining results are consistent. reproducible . for a batch of the stain seemed simple enough. so and DNA-specific; Bruce Moyer of ORNL ·s Chem istry Division readily agreed to prepare it. fo ll owing Gautier's 0 The ORNL stain is completely so luble in published procedure. As Gautier's publication water (unlike Gautier's stain) and leaves less warned, however. the stain proved to be difficult visible precipitate on the specimen (see to make and use and not reproducibly specific for electron micrographs on page 52): DNA. Moyer's interest and previous experience 0 The new stain has a longer shelf life. making investigating the chemistry of heavy-metal it more suitable for research use: and compounds. and the obvious need for a better. 0 Greater imaging contrast is provided by the more reliable. staining agent for DNA. subse­ new compound, even when tenfold lower quently led Moyer and Ada Olins to file a joint concentrations than usual are used. application for seed money to synthesize and chemically characterize an improved DNA­ The contrast between cells stained with a specific cytochemical stain. conventional agent and cells stained with the new Beginning with a thorough literature search. ORNL osm ium-ammine preparation is illustrated Moyer prepared a chart identifying several in the electron micrographs of rat liver ce ll s promising osmium-based compounds having shown on page 52. The top cell. which was potential DNA-staining capabilities. With the prepared with a conventional heavy-metal agent. funding received. he was able to enlist the services strongly stained most cell ul ar structures. showing of an Oak Ridge Associated Universities little specificity. In the bottom cell. treated with Postgraduate Research Associate. Sook-H ui Kim, the newly developed osmium-ammine complex.

Number Three 1988 51 NEW DNA STAIN AIDS CELL STUDIES

After a half-dozen subse­ quent chemical syntheses of other osmium compounds by Kim gave products inferior to this successful mi xt ure, the project participants shifted their focus to separat­ ing and purifying the active staining ingred ient of the new reagent and optimizing procedures for its preparati on Electron micro­ and use. graphs of rat liver Efforts to chemically cells (bar length = characterize the new stain IJ.lm) stained with resulted in the separation of conventional three components, and there agent (top cell) and new ORNL may be others. Two of these agent (bottom components were purified cell). and identified by both their spectral and chemical proper­ ti es as nitrido-bridged osmium dimers (see figure on page 53). Although the exact arrangement of the atoms in these compounds remains to be determined, the key structural feature is thought to be a "backbone" consisti ng of two osmium atoms linked by a nitrido li gand (N'-. essentially a fully dissociated ammonia molecule). The diagrams shown here illustrate the hypothetical structure of these dimers. Dimer I has a structural formula that has been known for nearly 35 years. but dimer 2 is a new osm ium compound. When only the DNA-containing structures are strongly dissolved in water, both dimers initially behave as stained. enabling cytologists to differentiate these salts, in that the chloride ions shown outside the from other ce ll inclusions. The large arrows point brackets dissociate immediately. Over a peri od of toward the nucleolus of each cell. Almost total days, however, the chl oride ions sho wn bonded staining occurs with the conventional agent (top), di rectly to the osmi um atoms also dissociate and but the ORNL agent (bottom) stains only the are replaced by water molecu les, in a process DNA-containing chromatin structures in the called aq uat ion. Each of the osmium complexes in nucleolus. the new set thus formed has its own indi vidual

52 Oak Ridge National Laboratory REVIEW properties, which have been only partially controls the expression of genes. For this work. explored. she was named one of six 1988 winners of th e Ironically. though dimer 1 was exactly the University of Tennessee Chancell or's Awards for compound sought by Kim and Moyer when they Research and C reative Achievement. Ada Olins planned the synthesis of the new stain, Allison's has also been an Eleanor Roosevelt International testing of the purified Cancer Fellow, and compound showed Donald O li ns has that it exhibited no been an American staining ability. Cancer Society Purified dimer 2. as Nf-i:J Nf-i:J Scholar. Both well as the aquated I / NH3 I /NH3 Olinses worked, in Cl - Os 'J .. N ~ Os'::_ Cl derivatives of both Cl 3 1 1987. as vis iting sci­ I dimers, also fai led to ~ 31 ;H31 entists at the Pasteur NH NH stain DNA. Unfortu­ r 3 3 1 Institute in France Possible nately, funding for the and at the Max structures of the work expired before Planck Institute for n itrido-bridged osmium dimers. the active staining NH:J Nf-i:J Biophysical Chemis­ component (or combi­ I /NH3 I /NH3 try in the Federal Re­ nation of components) Cl - Os ~-- N~ Os'::_ Cl Cl 2 2 public of Germany. could be detem1ined. N~ ~ el l Although a patent But, the project NH3 NH3 1 disclosure has been achieved its goal of l fil ed for the newly developing a more developed staining reliable cytochemical agent, Moyer sees stain and has contrib- li ttle immediate uted valuable information about the inorganic economic incentive for commercial production of chemistry of osmium compounds. the new stain. However, both Moyer and the O linses believe it will be a valuable aid to the Visualizing Cell Structures basic biological research that is resolving the chromosomal structures wi thin living cells. The Olinses are eager to continue their work Cell biologists have an equally fundamental using the new DNA-specific stain and their need for an improved ribonucleic acid-specific electron microscope tomography method to create (RNA-specific) stain. If funding can be obtained, a sophisticated three-dimensional reconstruction future research will be directed toward the of the assymetric DNA-containing cellul ar struc­ development of such a stain through the collabo­ tures. Ada Olins was recently honored for her re­ rative testing of various osmium complexes and search contributions toward visualizing the toward continuing the challenging chemical structure of chromatin, a complex of DNA and characterization work. =:I proteins that makes up the chromosomes and

Number Three 1988 53 I;~ 3: Awards and Appointments

Alvin W. Tri velpiece, Louis K. Mansur has been Walter P. Eatherly has been executive officer of the appointed to the editorial elected chairman of the American Associati on for the boards of Adranced Materials ASTM Committee C-5 on Advancement of Science and and Processes, Journal of Manufactured Carbon and former director of the Office Nuclear Materials. and Graph ite Products. a of Energy Research of the Journal of" Materials standards-writing committee. Department of Energy, has Engineering. Robert Hightower and been named director of Oak R. J. Norby has been W. Wilson Pitt, Jr. , have Ridge National Laboratory appointed to the editorial been named fellows of the and vice president of Martin review board of the journ al American Institute of Chemi­ Marietta Energy·systems, Inc. Tree Physiology. cal Engineers. Raymond S. Wiltshire, Linda Cain served as Chester R. Richmond has 0. B. Morgan former ORNL executi ve president of th e 39th Interna­ been elected to the Board of director, has been named vice­ tional Science and Engineer­ Directors of the National president fo r Computer-Aided ing Fair in Knoxville, the Council on Radiation Protec­ Producti vi ty of Martin largest such fa ir ever held. tion and Measurements. Marietta Corporation. On Sep­ Glenn W. Suter II has M. B. Adams has been tember I, 1988, he assumed received th e annual Scientific elected to the Executive hi s new positi on at corporate Achievement Award of Steering Committee of the Air headquarters in Bethesda, ORNL's Environmental Pollution Workshop, an Maryland. Replacing him as Sciences Division. an nual international meeting ORNL executive director is R. Julian Preston has been of scientists and policy 0. B. Morgan, forme rl y named a member of the managers in terested in air di rector of ORNL 's Fusion Toxicology Study Section of quality issues. Energy Di vision. John the National Institutes of Sheffield, former associ ate Heal th . Martin Mari etta Energy John Sheffield director of the Fusion Energy Diana Popp has been in vited Systems, Inc., honored its Division, has been named its to write an article on employees on May 20, 1988. director. .. Lymphocytic Subsets and at the fourth annual dinner and Martin Marietta Energy Abnormali ties in Retrovirally awards presentation program Systems, Inc., has conferred Infected Mice" fo r publication at the Hyatt Regency Hotel in its highest honor, the rank of by CRC Press in Crirical Knoxv ill e. More than 130 senior corporate fellow, on ReFiews in Immunology. ORNL employees or Energy Sheldon Datz. Only three ORNL women employees Systems employees working other Energy Systems who received 1988 awards on ORNL projects received em pl oyees hold this from the East Tennessee awards. The engineer of the di stinction. Chapter of the Association for year was Ray L. Johnson; the Dale D. Huff has been Women in Science (A WIS­ in ventor of the year was named head of the new ETC) are Elizabeth Peelle, Terry Tiegs; the author of the Environmental Engineering Distingui shed Leader and year was Mark Rasolt; and and Hydrology Section in Advocate A ward ; Kimiko the manager of the year was Sheldon Datz ORNL's Environmental Bowman, Di stinguished Paul Haubenreich, who is Sciences Division. Achievements in Science now secretary of the Interna­ Tuan Vo-Dinh has been Award; Ellen D. Smith and tional Thermonuclear Experi­ named to the editorial board Lynn L. Wright , O utstanding mental Reactor Council of the of Applied SpeClroscopy. A WIS-ETC Service Award. International Atomic Energy John F. Cooke has been Ray C. Hudson has been Agency in Vienna, Austria. elected a fel low of the named ORNL Standards They are among the five American Physical Society. Coordinator. Energy Systems employees

54 Oak Ridge National Laboratory REVIEW who each rece ived Marti n deposition techniques; mechanics methods to Marietta Corporation's Terry N. Tiegs, for innova­ evaluate pressure vessel em ­ highest awa rd, the si lver tive approaches to the devel­ brittlement problems and to Thomas Jefferson Cup. opment of ceramic composites prov ide a so un d techn ical for high-temperature basis for projecting the Fourteen ORNL employees appli cations; Graydon L. vessel's remaining useful life; received the Inventor Award, Yoder, Jr., Jack E. Smith. David K. Christen, for wh ich recognizes innovative and David G. Thomas, for continued excellence in em pl oyee contributions to the in vention of a multilayer experimental investigations activities of Energy Systems. shi eld to protect orbiting that have led to a better The wi nn ers were Charles W. spacecraft against hyper­ understanding of regu lar and Forsberg, for the invention of ve loc ity impact by micromete­ high-temperature supercon­ the Process Inherent Ult imate orites, space debris. and ductors, which constitute a Ray L. Johnson Safety Boiling Water Reactor, orbi ti ng particles. technologically important a power reactor in which the class of materials; James M. safety systems contain no Thirty-eight Ene rgy Systems Corum, for outstand ing electronics or moving parts; employees work ing on ORNL leadership and contributions John B. Hayter and Herbert projects received Technical to international collaborati ve A. Mook, Jr., for innovative Ach ievement Awards. which efforts to develop a structural design and development of recognize excell ence of design methodology fo r improved neutron opti cal emp loyee contributions of a advanced high -temperature devices; Scott Hunter and scientific or engineering reactors; Cabell B. Finch and Loucas G. Christophorou, nature to th e activities of Lynn A. Boatner, for the fo r development of gas Energy Systems. The winners soluti on of technical problems mixtures that possess tempera­ were Kimiko 0. Bowman . and contributions to the devel­ ture-enhanced glow discharge for a lifelong contribution to opment of the first prome­ characteristics for use in the theory of statistics with thium laser; Michael R. Terry Tiegs repetitive pulsed-power emphasis on properties of Guerin, for pioneering the closing sw itches; Mark A. small sample estim ators and development and appl icati on Janney , for numerous computational statistics: of modern organ ic analytical innovative contributions to the Everett E. Bloom , for chemistry with in the health synthesis and processing of sustained leadership in the and environmental programs ceramic powders: Harold D. development of structural of the Department of Energy Kimrey, Jr. , and Terry L. materials for fusion power and many other agencies and White, for the development of systems; David 0. Campbell, instituti ons: Richard G. a microwave furnace and for his experti se, which has Haire and John K. Gibson, insulation concept for merited worldwide recogni­ for the first measurement of processing ceramics; tion. and his many significant the cohesive energy of fer­ Vinod K. Sikka, for unusual contributions to chemical mium , element I 00. thus creati vity in devising new and aspects of nuclear fuel es tab li sh ing the divalent improved procedures in reprocessing. separations nature of the bonding in materials processing, for de­ chemis try , reactor chemistry, fermi um metal; Steven P. Mark Rasolt velopment of new materials. and nuclear waste Hirshman, for long-term and for innovative we lding management; Richard D. contributi ons to creation of a techniques; David P. Stinton, Cheverton, John G. Merkle, theoretical basis for analyzing for numerous contributions to and Randy K. Nanstad. for fl uid equi li brium and colli ­ the development of advanced outstandi ng leadership in the sional particle and heat materials through the development and application transport in complex three­ application of chemical va por of advanced fracture- di mensional magnetic

Number Three 1988 55 AWARDS AND APPOINTMENTS

systems; Wayne A. developmental biology, and professional field. The Houlberg, for his long-term radiation biology; Harold G. winners were C. Bottcher and contributions in development Smith, for enhancing the M. R. Strayer, Relativistic and validation of method for understanding of the marten­ Theory ofFermions and time-dependent modeling of sitic phase transformation in Classical Fields on a Colloca­ toroidal fusion plasma de­ lithium metal through tion Lattice; J. D. Budai, J. Z. vices; Ray L. Johnson, R. D. neutron-scattering experi­ Tischler, A. Habenschuss, Benson, M. J. Cole, B. E. ments; Cullie J. Sparks, Jr., G. E. Ice, and V. Elser, Nelson, J. A. White, K. K. for development of ORNL's X-Ray Diffraction Study of Chipley, D. J. Taylor, G. H. Synchrotron X-Ray Facility at Phason Strain Field in Henkel, M. J. Saltmarsh, the National Synchrotron Oriented Icosahedral Alumi­ and J. L. Yarber, for leader­ Light Source and for leader­ num-Manganese; T. A. Kimiko Bowman ship in completion of the ship in the application of Carlson, P. Gerard, M. 0. intricate final assembly of the synchrotron X-ray radiation in Krause, F. A. Grimm, and Advanced Toroidal Facility materials research; Dennis J. B. P. Pullen, Photoelectron within the required tolerances Strickler, for his pioneering Dynamics of the Valence and within budget; Kenneth contributions to defining the Shells of Benzene as a Liu, for the development of a poloidal magnetics design of Function of Photon Energy; high-temperature mechanical elongated divertor plasmas in L. G. Christophorou, S. R. properties test system that the Compact Ignition Toka­ Hunter, L.A. Pinnaduwage, allows the determination of mak in collaboration with J. G. Carter, A. A. Christo­ tensile properties of brittle Princeton Plasma Physics doulides, and S. M. Spyrou, ceramic materials; Richard Laboratory scientists; David Optically Enhanced Electron Lorenz, for significant G. Thomas, Jack E. Smith, Attachment; S. A. David, technical achievements, from and Graydon Yoder, Jr., for J. M. Vitek, and T. L. 1961 through 1987, associated the conception and demonstra­ Hebble, Effect of Rapid with determination of fission­ tion of a kinetic energy/laser Solidification on Stainless Richard Cheverton product source terms from shield for space platforms Steel Weld Metal Microstruc­ nuclear reactor accidents; with effectiveness equal to tures and Its Implications on Samuel B. McLaughlin, Jr., that of conventional concepts the Schoeffler Diagram; D. C. for outstanding research but with only I 0% the weight; Gregory, L. J. Wang, F. W. contributions to understanding and Man H. Yoo, fo r his out­ Meyer, and K. Rinn, Elec­ the physiological responses of standing contribution to the tron-Impact Ionization of Iron forests to air pollution stress; fundamental understanding of Ions: Fe 11 +, Fe 13+, and Fe 15+; Robert E. Mesmer, for micro-mechanisms re pon­ J. A. Horton and M. K. sustained scientific contribu­ sible for thermal strengthening Miller, Atom Probe Analysis tions and leadership in basic and embrittlement of struc­ of Grain Boundaries in research with high tempera­ tural metals and alloys, Rapidly Solidified Ni~l ; tures and pressures, in including ordered intermetal­ C. H. Johnson, D. J. Horen, aqueous chemistry, and in lic compounds. and C. Mahaux, Unified De­ geochemistry; Michael K. scription of the Neutron-208Pb David Christen Miller, for the design and Sixty-five ORNL employees Mean Field Between -20 and development of ORNL's atom received a Publication Award, +165 MeV from the probe and for its application to which recognizes superior Dispersion Relation important questions in employee performance in the Constraint; F. W. Larimer, materials science; Liane B. authorship of a paper, E. H. Lee, R. J. Mural, T. S. Russell for a brilliant research technical article, or book that Soper, and F. C. Hartman, career in mammalian genetics, represents a significant I ntersubunit Location of the reproductive and advance in the author's Active Site of Ribulose-

56 Oak Ridge National Laboratory REVIEW Bisphosphate Carboxylase! Fourteen ORNL employees Stallions, for three consecu­ Oxygenase as Determined by (or Energy Systems employ­ tive years of innovative In Vi1•o Hybridi:ation of Site­ ees involved in ORNL leadership and strong manage­ Directed Mutants; S. H. Liu, projects) received an Opera­ ment skills that led hi Electronic Polaron Effects in tional Performance A ward, Safeguards and Security Heavy-Electron Materials; which recognizes outstanding Department and ORNL to D. H. Lowndes, S. J. exemplary performance in achieve "superior" ratings Pennycook, G. E. Jellison, management, business. from DOE on the Nuclear Jr., S. P. Withrow, and D. N. personnel, manufacturing, and Materials Control and Mashburn, Solidification of other similar functions. The Accountability function; Highly Undercooled Liquid winners were Sheila T. Victor J. Tennery, for Silicon Produced by Pulsed Brooks, for outstanding leadership in expanding high­ Laser Melting of Jon­ contributions in Performance temperature materials research Vic Tennery Implanted Amorphous Silicon : Improvement Projects, special at ORNL, especially in Time-Resolved and Micro­ assignments at ORNL, and development of the new High­ structural Studies; P. K. long-time exemplary perform­ Temperature Materials Mioduszewski, R. C. Isler, ance as a senior secretary; Laboratory and in managing J. E. Simpkins, P. H. Paul N. Haubenreich, for his research in the High-Tempera­ Edmonds, E. A. Lazarus, personal commitment and ture Materials Section of the C. H. Ma, and outstanding contributions to Metals and Ceramics Divi­ M. Murakami, Impro vement the international Large Coil sion; Christine S. Travaglini, of Plasma Pe1jormance with Task for the development of for sustained outstanding Chromium Gettering in/SX­ fusion superconducting performance in the develop­ B; G. S. Painter and F. W. magnets; Lester J. King. for ment and administration of the Averill, Effects of Segregation leadership and coordination of Laboratory's capital equip­ on Grain-Boundary Cohesion: extensive efforts to provide ment, budgets, overhead, A Density-Functional Cluster safety documentation and material, and cost projections; Christine Travaglini Model of Boron and Sulfur in technical requirements for and Frederick W. Young, Nickel; M. Rasolt, Supercon­ successful startup of the Jr., for outstanding susta ined ductivity in High Magnetic TURF Californium Facility; management and program Fields; R. D. Spence and Sterling A. Meacham, coordination in the solid state A. L. Wright, The E. Craig Bradley. Bruce M. sciences. Importance of Fission WinchelL and Donald W. Production/Aerosol Interac­ Jared, for planning, develop­ Felicia M. Foust received a tions in Reactor Accident ing trategies for, and leading Community Service Award, Calculations; D. A. Spong, the effort to commerciali ze which recognizes outstanding D. J. Sigmar, K. T. Tsang, the Advanced Servo­ and noteworthy performance J. J . Ramos, D. E. Hastings, manipulator; Morris H. by Energy Systems employees and W. A. Cooper, Effects of Slabbekorn, Kenneth S. engaged in voluntary, uncom­ Alpha Populations on Davis, and R. Bruce pensated activities (social, Tokamak Ballooning Stability; Johnston, for development civic, or governmental) that T. Vo-Dinh, B. J. Tromberg, and sustained improvement in provide significant benefit to Fred Young G. D. Griffin, K. R. Graphic Arts capabilities that the community. She was Ambrose, M. J. Sepaniak, have resulted in a state-of-the­ honored for 30 years of and E. M. Gardenhire, art operation and consistently dedicated involvement in Antibody-Based Fiberoptics high placement in national volunteer work with the Biosensor for the Carcinogen competition with private elderly and with schools, Benzo(a)pyrene. companies and governmental athletic programs, and other agencies: Donald R. youth activities. IC:!J

Number Three 1988 57 • ! ;]~:Books CHA OS: Making a New Science by James Gleick, Viking Press, New York, 1987,352 pages. Reviewed by N. B. "Woody" Gove, Data Systems Research and Development, Martin Marietta Energy Systems, Inc.

Is there order in chao ? Can a system that scienti ts sometimes find other terms, such as exhibits chaotic, apparently unpredictable "studies of nonlinear phenomena" or "fractal behavior actually be following a simple set of geometry" more acceptable. The terms previou ly rules? Classical science has traditionally been mentioned are increasingly used to de cribe "The most concerned with describing systematic, orderly various manifestations of this new science, which behavior-such as the rotation of planets about has been called a science of process rather than ardent chaos the sun or the tidy linear relation hips of Euclid­ state. advocates ian geometry. When confronted with di sorder in Gleick points out that, in its descriptions of the believe that the natural world, such as the turbulence of ocean global nature of y terns, chaos science has cut or wind currents, fluctuations of wildlife popula­ across the lines of scientific disciplines and 20th-century tions, or oscillations of the heart, classical reversed the trend toward specialization that science will physicists and mathematicians have generally characterized science in the 1970s. In fields such been ignorant or, in some case , annoyed. But in as particle physics and molecular biology, be remem­ the 1970s, a few scientists around the world scientist at that time seemed to be learning more bered for just became interested, even concerned, about the and more about less and le . The empha is wa three things: inability of cience to address such problems. on analyzing the mal lest details of systems: Working in many separate fields, they began to quarks, chromosomes, or neurons. Chaos scientists relativity, think seriously about disorder, and to wonder are looking at the whole, seeking order and quantum whether there might be some universal principles universal behaviors within nature's randomnes connecting the various types of irregularity found and complexitie . The mo t ardent chaos advo­ mechanics, in nature. Gleick 's book pulls together the cates believe that 20th-century science will be and chaos." thinking of these "out-of-step" scientists who remembered for just three things: relativity, began looking at the natural world from a new quantum mechanic , and chaos. perspective, eeking to find some univer al order Many systems. both physical and mathemati­ within its many form of "chao ." cal, can be shown to have a very sen itive depend­ Mitchell Feigenbaum, a Los Alamos physicist ence on initial conditions. Even simple recipe for who thought about the turbulence in cloud generating numbers can lead to surprisingly formations and lightning path -watching them complicated patterns. Physicist Edward Lorenz, from airplane windows unti l his scientific travel working with one of the earliest computers at the privileges were suspended on grounds of over- Massachusetts Institute of Technology, discovered u e-was one of the early investigators of chaotic how a very mall perturbation can cause an phenomena. Almost simultaneously , a mathemati­ ultimately huge change when he hurriedly entered cian in Berkeley, California, a population some rounded-off numbers in a simple computer­ biologist at Princeton University, a geometer ized weather-forecasting program. The original working for IBM, and a French mathematician program, tored in the computer's memory, had studying liquid turbulence were also beginning to numbers with six decimal places; on a hurried day investigate the complex behaviors that can result Lorenz reentered the numbers, rounded to the first from relatively simple model systems. They three digits, assuming that such a small change began searching for some organizing principles or (one part in ten thousand) would cause a corre­ periodicity, using new techniques and the newly spondingly small change in the weather pattern available computing power to analyze their predicted by the computer. To his surprise, the apparently random data sets. New terms, such as pattern that emerged was grossly different from "dynamical ystems," "fractals," and "strange anything the program had previously predicted. attractors" began appearing in scientific literature. Intuitively. Lorenz realized that thi s effect was ow, nearly two decade later, thi fast-growing enormously unexpected and important. Ultimately, movement ha come to be known under the it meant that long-range weather forecasting is shorthand name of "chaos," although traditional impo ible, becau e the many small perturbations

58 Oak Ridge National Laboratory REVIEW occurring constantly in the real atmosphere can "stretched'' their previous scientific experience have enormously magnified and unpredictable and disciplines to study chaos. effects. He later described this phenomenon, Some looked for patterns in chaotic systems which he called the Butterfly Effect, in a paper and others found chaos, or nonlinear, unpredict­ • (presented in 1979 at an American Association for able behavior, within systems previously believed "After the Advancement of Science conference) entitled: to be quite orderly and predictable, even going so "Predictability: Does the Flap of a Butterfly's far as to proclaim that "the regularity Galileo saw reading Wings in Brazil Set Off a Toma.do inTexas?'.' is only an approximation." At low amplitudes of Chaos, you Lorenz applied his new way of thinking to any dynamic system, any extremely small will never develop some simple-looking equations describing nonlinearity can be-and often has been­ a hydrodynamic process called convection. He attributed to error and, for practical purposes look at the asked the question, "How can we calculate how appears to have no discernible effects, but " It is world in quickly a hot cup of coffee will cool?" If cream is there and it is measurable," and its effects are added, the swirls produced by convection cari be real, say the chaos investigators. quite the complicated, and Lorenz realized that traditional Robert May, an Australian physicist turned same way physics provided no means to calculate the biologist, studied a simple equation used for agazn.. " patterns of these convection currents or predict population studies, x(next) = rx( 1-x). He showed their cooling effects. Even though the set of that. for certain values of r, the population would convection equations Lorenz developed appear oscillate between two points. At higher values, a simple, they contain nonlinear terms and, as phenomenon now called "period-doubling" or Lorenz explained, "You think there must be a way "bifurcation" occurs, with population values to get around them. But you just can't." He reaching four times those expected, then eight, pointed out that it is very difficult to predict the and eventually chaos. According to Weinberg, temperature of the coffee one minute in advance, this type of behavior was seen earlier by E. Fermi although it is easy to predict it for one hour ahead, and S. Ulam in a nonlinear oscillator model. (At because it will eventually reach room temperature. first, I thought Gleick had missed this item, but Lorenz and other mathematicians and physicists it's actually mentioned in the notes at the back of went on to develop other systems in which our the book.) James Yorke, a University of Mary­ intuition-and classical science--can be proved land mathematician, proved that, if an oscillation wrong. Consider the behavior of such diverse involving three variables occurs in any one­ "systems" as waterwheels, population models, dimensional system, then chaotic behavior will investment curves, and static occurrences on a also occur. From his paper, "Period Three Implies telephone line-all of these could be expected to Chaos,'' an emerging field of science took its reach some steady state under certain conditions. name. But Lorenz and others found that they do not; In the 1970s, Benoit Mandelbrot, an IBM instead, they show a strange sort of regularity mathematician, realized that Euclidian measure­ witrun irregularity-always changing in a nonlin­ ments could not capture the essence of irregular ear, unpredictable fashion, yet always remaining shapes. He puzzled over problems such as how to within certain bounds and showing unusual islands describe the dimensions of a ball of twine. With of periodic stability within the chaotic whole. his access to what Gleick describes as the "high­ Gleick describes, in a very readable way, the speed idiocy" of IBM's considerable computing new wave of research approaches that some resources, Mandelbrot developed a new geo­ (including former ORNL Director Alvin Wein­ metrical concept called fractals, characterized by berg) regard as a "revolution" or "paradigm shift" "scale symmetry." If a rough surface is magni­ in science comparable to that of relativity or fied, it may show both smooth and rough areas, quantum mechanics. The book also traces the and the "degree of roughness" usually depends on history of chaos studies and the people who left or the degree of magnification. A fractal surface,

Number Three 1988 59 RE: BOOKS

however, exhibits the same degree of roughness figures on the opposing page. The contrast or irregularity pattern at any scale of magnifica­ between quasiperiodic and chaotic motion for a tion. Clouds, feathers, cauliflowers, and volcanic classical two-node coupled oscillator model is plumes are examples of the fractal surfaces found illustrated in (Ia) and (lb). For both trajectories, in nature. Our human mind cannot visuali ze the the Hamiltonian is the same and is very similar to "Some of whole of such infinite self-embedding complexi­ that for the famous Henon-Heiles model, well­ ties, but mindless computers can create them by known to physicists. In chemical appli cation, it the new performing endless iterations of a mathemati­ shows the effect of coupling between two vibra­ computer cian's imagined form. Such a surface could even tional modes of a polyatomic molecule. Figures (theoretically) be infinite in area yet enclose a (2a) and (2b) represent the corresponding and mathe­ finite volume, and several mathematicians since quantum-mechanical probability densities for matics Mandelbrot have delighted in finding or devising these two trajectories (see Noid's discussion of methods such fractal figures, once they knew what to look chaos science in "Using a Supercomputer for for. Fractal geometry has been applied to many Theoretical Chemistry,"' on page 10 of this issue). described natural systems and problems that exhibit some Chaos researchers have shared the same in this book degree of scale symmetry, such as earthquake and hardships most revolutionaries have-lack of income distribution. tree shapes. the human fund ing, a scarcity of advisors. and difficulty in are also circulatory system, neural network , snowflakes, having their work accepted (and published), currently and galaxies. Christopher Scholz. a Columbia because frequently it does not fit the boundaries of University professor specializing in the form and any traditional discipline or match the goals of a being used structure of the earth, consider fractal geometry funding source. Chaos research remains controver­ at ORNL." indispensable to his work, calling it "a single sial, but it has gained respectability and proven its model that allows us to cope with the range of usefulness in many areas. For example, Ilya changing dimensions of the earth.'' Prigogine, a Nobel laureate chemist and author of In 1971 , David Ruelle, a Belgian mathematical a 1984 book, Order Out of Chaos: Man's New physicist, and Floris Takens, a Dutch mathemati­ Dialogue with Nature, organized a 1985 confer­ cian, published a paper, ''On the Nature of ence on chaos in economic . and a NATO Turbulence," which introduced the term "strange conference on fractal and nonfractal patterns in attractors" to describe the patterns of nonre­ physics was held in 1986. Los Alamos National peating oscillation around two or more points in Laboratory recently established a Center for phase space, leading to an appearance of chao . Nonlinear Studies, and Oak Ridge National The term caught on, and strange attractors oon Laboratory started a Center for Nonlinear became the subject of several conference . (A Phenomena that is now operated by the Unjversity good description of strange attractors is given by of Tennessee. The U.S. Defense Advanced Douglas Hofstadter in the November 1981 issue Research Projects Agency (DARPA) is funding a of Scientific American.) The strange attractors Georgia Institute of Technology project on "rule­ concept has been used in studies of phenomena based" algorithms for fractals, with wide-ranging ranging from galactic orbits to dripping faucets; applications-missile target recognition, pine Gleick says they are everywhere in nature, if we forestry, submarine design, the study of human know how to look for them. faces, flight simulation, and cinematic special Scientists have found the universal property effects (a discussion of the latter appears on they sought in chaos. In 1977 two physicists, page 28 of the February 1988 issue of Scientific Joseph Ford and Guilio Casata, organized the first American). Chaos theory has also appeared in the chaos conference in Como, Italy. ORNL chemist local press. in an economics article published by Don Noid attended this conference. Then a the Knoxville News-Sentinel on February 14, 1988. postdoctoral researcher at the University of Some of the new computer and mathematics Illinois, he presented a paper on polyatomic methods described in thi book are also currently dissociation by infrared heating that included the being used at ORNL. For example, Sam Liu, Ted

60 Oak Ridge National Laboratory REVIEW These figures, some of the first examples of quantum­ mechanical X AXIS representation of a chaotic system , ''' were presented by • I Don Noid at the first chaos science conference at Como, Italy ,

~ 0 in 1977.

-2

- 4 -·

-·~------~ - 3 -2

Kaplan, and Len Grey used fractals in studying the Certainl y Gleick presents a fascinating wealth of interface betwee n a metal and an electrolyte. Mark material on a wide range of chaos problems in Raso lt is investigating the turbulence caused by natural systems (several pages on cardiac models, large optical exc itations in semiconductors. Don for example) and on th e sc ientists pioneering in Noid does research on chao tic behavior in this fie ld. Because Gleick. a New York Tim es polymers. reporter. is writing for the public, he avoids The reader may wonder how many of the technical terms and equations and does not ·'new·· results of chaos theory are really new. It is attempt mathematical or physical preci sion in hi s probably impossible to trace the in tellectual discussions. Extensive refere nces are given for heritage completely, but Gleick does describe technical readers who want to investigate further. earlier related works and adds 20 pages of notes Gleick conveys a sense of excitement in the on source s. Id eas on unpred ictability and sensitiv­ promise or hope that a new scientific era is ity to initial conditions are traced as far back as beginn in g and that everyday phenomena may be , Poincare '. Fractals are related to a 60-year-old if not predictable. at least better understood. He class of shapes called Julia sets (after French suggests that the new science may revitalize and mathematician Gaston Julia), and two earl ie r reunite oth er disc iplines. conjuring up a picture of mathemati cal creations of infinite complexity. young renegades overthrowing the established called '"Cantor dust'" and the '"Koch snowflake ... order to set up a new utopia. Even if hi s expecta­ are also desc ribed. tions are overdrawn and chaos re search causes "After reading Chaos. yo u will never look at on ly a little chaos in sc ience , thi s book is impor­ the world in quite the same way again ... states the tant--and fasc in ati ng--reading. =:I book jacket. This assertion may well be true.

Number Three 1988 61 :R&D Updates

Trivelpiece Is ORNL'sNew Director

ak Ridge National Laboratory will securing a Laboratory director with such out­ have a new director on January I. tanding background and experience. More than 1989. On September 30. 1988, 147 candidates were nominated or applied for the Oformer ORNL director Herman Postma directorship. announced the appointment of Alvin W. Acting Director Alex Zucker was commended Trivelpiece as ORNL director. Trivelpiece will by Postma for his effective leadership ince early also be a vice president of Martin Marietta February in what Po tma termed some ''very Energy Systems. Inc .. and will report to Postma. difficult times for scientific management." Zucker Energy Systems senior vice president. will continue to serve as director during the In hi s formal announcement to the Laboratory's transition period through the rest of this year. research and supervisory staff, Postma said. Postma pointed out that Trivelpiece is the fir t "Dr. Trivelpiece brings extensive academic, ORNL director appointed from outside the industrial, governmental, scientific. and adminis­ Laboratory and the first to be selected by an trative experience to this job, with a ound record internal search committee. He is, however, the of scientific and engineering accomplishments.'' second plasma physicist (Postma being the first) Trivelpiece. executive officer of the American and the second Alvin (Alvin Weinberg being the Association for the Advancement of Science. is first) to be named director. perhaps best known at ORNL for hi s work as Trivelpiece is also a journeyman electrician in director of the Office of Energy Research in the good standing, a certified flight instructor, and an Department of Energy from 1981 to 1987 and. avid marathoner. He peaks·fluent Dutch (as from 1973 to 1975. as assistant research director Postma doe ) and rides a unicycle (a feat Postma for the U.S. Atomic Energy Commi sion·s has tried and pronounced impossible). Postma Division of Controlled Thermonuclear Research. quoted Trivelpiece a saying that he feels he ha Both a Fulbright and a Guggenheim scholar, he moved ''too far from where science i done'' and has a doctorate in electrical engineering from the expect to enjoy being "back where important California Institute of Technology and has taught things in science happen." electrical engineering at the University of Califor­ nia at Berkeley and physics at the University of Two HHIRF Devices Maryland. His indu trial experience include executi ve positions with Science Applications. Begin Operation Inc .. and Maxwell Laboratories, Inc. Two new devices have recently come on line at In hi s informal remarks to the staff. Postma the Holifield Heavy Ion Research Facility praised the "thorough and very professional" (HHIRF). The Close-Packed Array, which packs work of the ORNL Search Committee. headed by 21 Compton-suppressed gennanium gamma-ray ORNL Associate Director Murray Rosenthal. in detectors around a central core is the re ult of a

62 Oak Ridge National Laboratory REVIEW decayed by positron emis­ sion to excited-state forms of • gold-191. The results ORNL researchers Noah Johnson provided information about (ri ght) and I. Y. energy levels in the mercury- Lee, architects of 191 nucleus. The NOF is the Compton­ unique in the United States Suppression­ in its ability to characterize Spectrometer nuclear decay modes and test system, watch as nuclear theories. the Close-Packed Array is fitted nearly $2 million collaborative effort involvino Tritium Pellet Injector Works around the scientists from ORNL, the University of Tennes-"' reaction chamber. see, and Vanderbilt University. The special Refueling fusion plasmas with pellets of frozen detector elements (see photo above) were devel­ deuterium, a heavy hydrogen isotope, has been oped in cooperation with EG&G ORTEC. demonstrated by ORNL in the 1980s at several One of the interesting uses of this device at the tokamaks. ORNL injectors ha ve HHIRF will be the search for super-defom1ed accelerated deuterium pellets into the plasmas of nuclei-heavy nuclear systems that, at high the Tokamak Fusion Test Reactor at Princeton angular rotation, assume a very elongated shape. Plasma Physics Laboratory and the Joint Euro­ Jim Ball , director of ORNL's Physics Division, pean Tokamak at Culham Laboratory in England. says these have been identified in only a few Now, ORNL has demonstrated for the first time cases, but some theorists believe they may occur that pellets of another frozen heavy hydrogen iso­ more commonly than now thought and that hyper­ tope-tritium--can be made and acce lerated. This milestone brings the fusion community deformed nuclei m;:~y be found with this powerful new detector array. closer to the ach ievement of deuterium-tritium Holifield's second new addition is the Nuclear plasmas. which are more likel y than deuterium Orientation Facility, recently added to the UNI­ plasmas to be used first in commercial fusion SOR Isotope Separator there. Ball says the NOF is power devices because they require less the most powerful instrument of its type in the world. With this advanced instrument, scientists are able to determine all angular distribution coefficients of the gamma-ray decay • from targeted molecules. The unique UNISOR staff, device for the study of nuclear along with faculty structure is operated by ORNL as a and graduate national user facility and is funded by students from Oak Ridge Associated Universities. eight universities This first NOF experiment involved who participated in a total of 20 researchers, including 3 the first successful on-line international research associates. 3 experiment, are UNISOR staff members, and 6 faculty shown at HHIRF's members along with 8 graduate new Nuclear students from 8 universities. In the Orientation experiment, the NOF magnetically Facility. oriented mercury-191 nuclei , which

Number Three 1988 63 R&D UPDATES

demanding con­ dition . Deu­ Larry Allard uses terium is readily the JEOL 4000EX transmission available from electron seawater, and microscope to tritium can be characterize high­ bred by neutron temperature bombardment of superconducting a lithium blanket materials. Th is in a fusion ultrahigh­ reactor. resolution In May 1988, microscope was the Tritium recently installed Proof-of­ in ORNL's High Temperature Principle (TPOP) Materials Pellet Injector, Laboratory. whichORNL designed and built, formed and accelerated pellets of frozen tritium at the Tritium System Test Assembly at Los Alamos National Labora­ tory. The TPOP device injected pellets (4 mm long and 4 mm in diameter) into the plasma at veloci ties up to 1200 m/ . have been active in high-temperature supercon­ Developers were P. W. Fisher, of the Chemical ductivity re earch. Technology Division, and members of the Plasma ··The pilot cen ters are a key element in the Technology Group of the Fusion Energy Superconductivity Initi ative announ ced by Division. President Reagan last year:· Secretary of Energy John S. Herrington sa id . They are designed to DOE Superconductivity speed up the con trac ting and project approval Pilot Center at ORNL process and th e transfer of patent ri ghts and other intell ectual property from the national laboratories One of three DOE Superconducti vi ty Pil ot to industry. Centers has bee n established at OR L. The The director of the ORNL pilot center i · centers will collaborate with industri es in Anthony C. Schaffhauser. Davi d Christen is advancing the technol ogy of hi gh-temperature sc ientific coordinator and Loui se Dunlap is superconducti vity. manager of the industry partnership program at the The other two pilot centers are located at Los ce nter. So far a number of pri vate companies. Alamo ati onal Laboratory and Argonne including AT&T Bell and IBM. have expressed National Laboratory. All three DOE laboratories interest in using ORNL"s fac ilities to characterize high-tem perature superconducting materials. c:::!J

64 Oak Ridge National Laboratory REVIEW : Technical Highlights Hood River Conservation Project Attracts Participants, Saves Energy

cently completed project in th e The results showed that an overwhelming Pac ific Northwest involving work by majority of utility customers are willing to ORNL researchers suggests that participate in an energy-saving program. that energyA conservation is an important .. so urce .. of cooperati on can be ach ieved between groups and elec tri ci ty . institutions that may oth erwise be at odds. and • The Hood Ri ver Conse rv ati on Project (H RCP) th at. adjusting for climate, th e Hood Ri ver "The results in Oregon has shown that home energy conserva­ community had the lowest space-heating con­ ti on measures-when app li ed systematically and sumption ever recorded follow ing a retrofi t showed that th oroughl y-can be a legi tim ate. reliable source of program. The resu lt s have been used by the an over­ power fo r pub li c and private utilities. ··conserva­ Bonneville Power Admi ni strat ion and the North ­ tion can be a cost-effective energy so urce:· says west Power Planning Council to design and plan whelming Eri c Hirst of ORNL, .. because it often cos ts less to future conservation programs. majority save electric ity th an it does to produce it. Thus. The HRCP spent about $4400 per household on of utility it hel ps el iminate the need to build expensive weatheri zation measures. As a res ult of these generating plants." measures. th e weatheri zed homes each reduced customers The five -year. $21-m ill io n HR CP ··turned out to electricity use by abou t 15 0r of th e amount are willing be a remarkably popular res ident ial weatheri zation consumed before weatherizati on (an average program. th ank s to extensive word-of-mou th savi ng of 2600 kWh per year). The savings would to partici­ communication among Hood Ri ve r residents and have been greater if the region had less unem­ pate in an th e project's offer of free weatheriza ti on ... says ployment and if many of the res idents had not Hirst. bee n acc ustomed to burning wood for space energy­ "About 90% of th e 3500 eli gi bl e households heat ing and taking oth er measures to keep electric savtng rece ived free energy aud its and in sta ll ation of bi lis down in response to the dramatic price rises program. water heater wraps. ceiling in sul ation. storm of the 1970s. windows. ca ul king. door wea therstri pp ing. duct The success of the project was partly th e resu lt insul ati on, and other measures to reduce consump­ of cooperation among environmentali sts. gove rn ­ tion of electricity for water heating and space ment officials. and util ity managers. It was heatin g. Thi s parti ci pati on rate is far higher than proposed by a private env ironmental protection th at usually ach ieved in conserva ti on programs:· group. th e Natural Resources Defense Coun ci l: Hirst points out. funded by a fe deral agency. the Bonn ev ille Power Most of the nonparticipants were owners of Ad mini stration: and operated by a private utilit y. si ngle-fami ly home s and earn ed higher-than­ the Pacific Power and Light Com pany in coopera­ average incomes. The unu sual ly la rge response to tion with the Hood River Electri c Cooperative. th e program yielded a ri ch harvest of inform ati on .. The questi ons behind the project were ' Is on the tangible benefits of conservation to a uti lity. conservation a true energy resource?' and ·can The projec t had another importan t component utilities rely on it?' From what we have learn ed besides weatherizati on: research and data collec­ from the Hood Ri ver projec t. the answers are tion and anal ys is. Data co ll ected--on the HRCP ·yes.'., said Hirst. who headed the evaluati on of from 1982 through 1986 were analyzed by th e the project fo r the Decision Systems Research Dec ision System s Research Section in ORNL ·s Sect ion. Much of th e analys is was done by group Energy Divi sion. leader Marilyn Brown and Rick Goel tz.

Number Three 1988 65 TECHNICAL HIGHLIGHTS

"Conservation measures in the home can keep consumers just as warm in winter and just as cool in summer as i can a coal-fired power plant,'' aid Contractors in stall Goeltz. "The difference is that storm windows on building a new plant could drive up a Hood River electricity rates to cover the capital house. inve tment and interest co ts. Our analysis shows that conservation can be a reliable, valid source of energy becau e it reduces the need for power." The extensive use of conservation as an energy resource is less co tly in the long run, says Brown. "The utility can buy conservation in small amount , see what works and what doesn't, and expand or reduce the program as needed. On the other hand, ordering a new nuclear power plant is a major, difficult-to-reverse commitment requiring $2 billion, I 0 years of con truction and testing, and pecial mea ure to satisfy the public and government regulators." The re earch and analysis provided by the Decision Systems Research Section and others began a year before the program started in 1983 and continued more than a year after con ervation measures were installed in 1985. According to Hirst, th is work was vital to the program's goal of determining the appropriate role of utilities in securing conservation resources to aid in regional energy planning. "ORNL was asked to evaluate the HRCP," explained Hirst, "because we have a good reputation for analyzing energy-u e data, because we were viewed as unbiased as are ult of our track record in conducting similar evaluations for the Department of Energy and Bonneville, and because we are located outside the Pacific Northwest region."

66 Oak Ridge National Laboratory REVIEW How Acid Rain Can Cause nitrate ions are introduced to the soil through acid precipitation, some of the aluminum is stripped Decline in Tree Growth away from the soil particles and brought into the The decline in tree growth in the Great Smoky acid solution. The solution can carry the previ­ ously soil-bound aluminum into streams or make Mountains National Park may be a result of the "Tree-ring greater availability of aluminum to the trees it available for absorption by tree roots. because of the effects of acid rain and air pollution Bondietti says that the evidence suggests that evidence on oils, according to an ORNL researcher. decreasing industrial emissions of sulfates and shows that Tree-ring evidence howing that aluminum nitrates should lessen the release of aluminum level are higher in trees whose outside rings have from soil and foster increased tree growth. aluminum decreased in size was presented May 12, 1988, The aluminum studies build on research done levels are by Ernie Bondietti of ORNL 's Environmental by Baes and McLaughlin in 1984, when they Sciences Division at the 14th Annual Scientific sampled numerous trees in East Tennessee and higher in Research Meeting of the National Park Service North Carolina. They deciphered the metallic trees whose at the Smokies park headquarter . element content of tree-ring tissues by a tech­ outside rings "The trees are the silent monitors of changing nique called inductively coupled plasma optical levels of industrial air pollution," says Bondietti. emission spectroscopy. have ''Since the 1940s, many Appalachian trees, from They found that annual growth rings from decreased the red spruces of New Hampshire to the shortleaf shortleaf pine trees in the southeastern portions of . . '' pines of Tennessee, have experienced declines in the Smokies showed both suppressed growth and m szze ... radial tree growth and increases in aluminum increased iron content between 1863 and 1912. levels. The effect is particularly noticeable for During this period, large sulfur dioxide releases the past 25 years, when regional fossil fuel from the copper smelting 56 miles upwind at combustion emissions from industry have Copperhill, Tennessee, cau ed extensive mortality increased as much as 200%." of trees within about an 11-mile radius of the Re earch by Fred Baes and Samuel McLaughlin smelter. mJ of ORNL's Environmental Sciences Division has shown that aluminum concentrations in tree-ring tissue have risen since the 1940s. The ORNL researchers have also ob erved that the tissues of trees showing decreasing ring growth have a higher level of aluminum than calcium. "Our findings,'· says Bondietti, "support the results of U.S. Forest Service research indicating that aluminum in the soil inhibits the uptake of calcium by trees. This reduction in calcium uptake may inhibit tree growth." Acid rain is formed when sulfur oxides and nitric oxides, produced from fossil-fuel combus­ tion, react with atmospheric moisture to form acids (sulfates and nitrates). Since 1984. ORNL re earcher , including Dale Johnson, have been examining the idea that acid precipitation makes aluminum, iron, and other metals in soil more available to trees and other vegetation. In most soils, positively charged ions of alumi­ num are attached to negatively charged soil particle . When negatively charged sulfate and

Number Three 1988 67 I I~~~: Technology Transfer ORNL Alloy Licensed to Supplier of Aircraft Fasteners

gh-temperature, nickel aluminide Metallamics To Manufacture lloy developed at ORNL ha been licensed to one of the aircraft Nickel Aluminide Products Aindustry's leading suppliers of bolts and other Energy Systems has announced a licensing fasteners. agreement with a new Michigan company that will "Energy Energy Systems recently granted Yalley­ use the high-strength, high-temperature nickel Todeco the exclusive rights to upply aircraft Systems aluminide alloy developed by ORNL for its fasteners made of nickel aluminide to the private recently products. sector. Yalley-Todeco, located in Sylmar, The company, Metallamics, Inc., of Traverse granted California, i a division of The Lamson & City, Michigan, has acquired nonexclusive rights Sessions Company. Valley­ to the heat- and corrosion-resistant alloy for the Nickel al uminide, which becomes stronger at fabrication of industrial part that must withstand Todeco the hi gher temperatures, is six times as strong as extremely high temperatures. stainless steel at 1100°F (600°C). Unlike many exclusive Normally a brittle material, nickel aluminide materials commonly used in the aerospace rights to has been made ductile through a patented ORNL industry, it is not subject to stress-corrosion process. The alloy i described as "intermetallic" supply cracking. because it shares some characteristics of both Yalley-Todeco President Patrick Dansby aid aircraft nickel and aluminum. ickel aluminide has the hi s company anticipates that nickel aluminide unusual characteristic of increasi ng in strength at fasteners will permit the development of new, high­ higher temperature . At ll00°F (600°C), it i six performance fasteners for the high-temperature made of times as strong as stainless steel. operating environment of tomorrow 's fuel­ Robert McDonald, Metallamics president, says nickel efficient engines. "Manufacturing and testing hi s firm intends to use a proprietary powder programs to determine the engine capabilities of aluminide to metallurgical process for specialty applications new high-strength, high-temperature materials the private of the ni ckel aluminide, such as tool and dies, have been started, and Yalley-Todeco is working cutting devices, and tructural and working sector." closely in this endeavor with a major manufac­ components. turer of metal alloys," Dansby added. Energy Systems negotiated the license with Yalley-Todeco following a waiver of patent rights Licensed Ionization Source from DOE. Licensi ng royalties, under a DOE­ Is Key Component of approved formula, will be used to support other Explosives Sensor technology transfer activities. Four previous licenses have granted the right to manufacture A California company has agreed to make a and market nickel aluminide and to use the alloy multimillion-dollar inve tment over a several- year in large diesel engines, heating elements. cutting period to design and produce a commercial devices, and tools and die . version of an ORNL device that can "sniff out" Other potential applications of nickel aluminide minute traces of th e telltale vapors from chemicals include hi gh-performance jet engines, gas tur­ such as concealed expl osives. Licensing efforts bines, advanced heat engines, and heat exchang­ were led by Jon E. Soderstrom, technology appli­ ers in nuclear and coal-fired steam plants. cations director for Energy Systems.

68 Oak Ridge National Laboratory REVIEW The Finnigan Corporation, based in San Jose, Cali fornia, has been granted a license for an Atmospheric Sampling Glow Discharge Ionization Source which, when coupled with a mass spectrometer, can in stantly detect low-level atmospheric chemicals. Finnigan is a worldwide producer of a broad range of mass spectrometers, which are analytical devices used to identify and quantify elements and isotopes in samples. • The proposed device will detect and identify "We see this explosives at levels lower than one part per billion in less than a second by chemically identifying device as organic nitrogen-oxygen compounds that are the having basis of explosives such as TNT, plastiques, and nitroglycerine. application The key to this increased sensitivity is the novel zn many ion source developed by Gary Glish and Scott McLuckey of ORNL's Analytical Chemistry fields, not Division. This source i a simple and durable only instrument, capable of high-sensitivity, trouble­ explosives free operation for prolonged periods. "We see this device as having application in detection." many fie lds, not only explosives detection," Michael Story, a Finnigan co-founder, said. "It is suited for many low-level remote sensing appli ca­ tions, such as hazardous wastes and industrial work -site monitoring, plant fugitive emissions monitoring, personnel or security screening, and even military appli cations." rl:!J

Number Three 1988 69