RESEARCH

.ï.ilS3Së Γ Interlocking Designs in Reforming Study

(Number of runs: 12+6=18)

Circles are for first factorial ex­periment—sulfur content at three levels, temperature and space velocity at two levels each.

Squares are for second factorial experiment—sulfur content is constant at an intermediate level, while temperature is at three levels and space velocity at two

Factorial planes întersect at x's. Note that moste éxper îmenta l points are centered in the area of

Interlocked Factorials: Fewer Runs New experimental design keeps advantages of full factorial design, cuts number of test runs

X HE ;MOST DATA with the fewest tests —this is the problem engineers and scientists have to face when they plan experimental programs. Now a new design comes to their aid. Called the interlocking factorial design, it can cut the number of tests to a fraction of those needed for a full factorial design (all combinations of variables and levels). Yet it keeps all the advantages of the full design, according to its de­signers.

There are other methods, such as the fractional factorial or the composite de­sign, to reduce the number of tests. But results may not meet the demands of engineering studies.

In the interlocking factorial design, you set up full factorial experiments using only two or three variables in each set, Basil J. Reitzer of Illinois In­stitute of Technology and John A. Brooks of Standard Oil ( Ind.) told the Division of Industrial and Engineering Chemistry at the San Francisco meet­ing of the AMERICAN C H E M I C A L SO­

CIETY. Then by choosing one level of

variables in set Β somewhere between the levels of those variables you chose in set A, you interlock the two facto­rial experiments and cover the full range of conditions including interac­tion, they say.

For example, if you have three vari­ables to be tested at four levels, you need 4 3 or 64 tests for a full factorial design. But if you break this up into two interlocking factorials, both of three variables at two levels, you need only 23 X 2 or 16 tests to get the same amount of information, Reitzer claims.

The only special requirements in setting up the experiments: Each vari­able must be varied independently of the others; levels at which each vari­able is set should follow some progres­sion such as arithmetic, geometric, or logarithmic.

• Five Designs. To prove out their method, Reitzer and Brooks set up a catalytic reforming study, investigating effects of five variables on octane num­ber. To do this with a full factorial de­sign having an average of four levels

per variable would take some 1000 runs, Reitzer says. Yet his interlocking factorial design cuts that to 5 1 .

Here is how they did it, having as variables temperature, pressure, space velocity, sulfur content, and hydrogen flow rate:

• Two factorial experiments give data on effects and interactions of tempera­ture, space velocity, and sulfur content, with pressure and hydrogen ra te held constant. In one factorial, they used three sulfur contents, two tempera­tures, and two space velocities; in the other, three temperatures (two of which were the same as in the other set) and two space velocities with sul­fur held constant (at an intermediate level). The result: 18 runs, covering four levels each of space velocity and sulfur content and three of temperature.

• A third factorial finds the effects of pressure, hydrogen rate, and tem­perature, takes 16 runs. Temperatures in this set are at levels intermediate to those in the first two experiments; hence the three factorials interlock.

• The fourth factorial ties together the first three by varying pressure and hydrogen rate at three levels each, takes nine runs. By using results from it, Reitzer says, they can relate the data from the third factorial to the

4 0 C & Ε Ν M A Y 5, 1958

standard conditions of pressure and hydrogen rate they used in the first two sets.

• The fifth factorial finds interactions among pressure, sulfur content, and hydrogen rate to complete the mathe­matical picture of catalytic reforming. Reitzer used two levels each of these three variables in this set—eight runs.

T h e grand total—51 runs. Engineers can use interlocking fac­

torials in many process studies, Reitzer feels. Any time you have many con­trolling variables in a process and a general idea in advance of the range of variables for best performance, this de­sign will pay off. he concludes.

Almost Synthetic Cellulose Mung bean seedlings lead to polysaccharides closely resembling cellulose

SYNTHETIC CELLULOSE is closer to reality, thanks to some recent experi­mentation. Three University of Cali­fornia scientists used mung bean seed­lings (and other plants) with carbon-14-labeled uridine diphosphate glucose. While t h e hoped for synthetic cellulose was not achieved, a polysaccharide was produced which closely resembled cellulose.

UC's D . S. Feingold and his cowork­ers told the ACS Division of Biological Chemistry at San Francisco that a solu­ble enzyme preparation obtained from the seedlings catalyzed the formation of a polysaccharide consisting of D-glucose units; uridine diphosphate was liberated. Partial hydrolysis of the polymer with acid gave a series of oligo­saccharides, which showed a chromato graphic pattern similar to those pro­duced b y partial hydrolysis of the poly­saccharide laminarin. Using this and other analytical data, Feingold feels justified in concluding that the D-glu-cose residues in the polysaccharide are joined by β-1,3 glycosidic linkages, similar but not the same as the βΊΑ linked oligosaccharides prepared from cellulose.

Feingold notes that the California group's findings that plant extracts can synthesize a β-1,3 linked glucan points to other questions: Is the glucan nor­mally present in plants? Do such poly­saccharides exist in higher plants? Or does the β-1,3 glucan trans-glucosylase occupy a similar position to invertase (present in plant extracts, but seem­

ingly inactive in the intact plant)? These are questions for future research.

Ultra-High Pressures High pressures at high tem­peratures now possible in tetrahedral anvil device

A NEW HIGH PRESSURE DEVICE h a s been designed which will enable scien­tists to work at ultra-high pressures and at high temperatures simultaneously. In the past, it was relatively easy to provide the energy required for chemi­cal transformations (10,000 to 100,000 calories per mole) by heating the sys­tem. Now, pressure ranging to 200,000 atmospheres can provide comparable energies.

But more important, the high tem­perature and pressure will be available with this device at the same time. Why is this of interest in high pressure studies? Generally, increased pres­sure retards chemical reactions, and in­creased temperatures almost always have t he opposite effect. Therefore, for chemical reactions to occur at high pressures in a reasonable length of time, it is desirable to operate at high tem­peratures also.

H. Tracy Hall of Brigham Young University told about his tetrahedral anvil design high pressure apparatus before the Division of Physical and In­organic Chemistry at the ACS meeting in San Francisco. This equipment can develop 100,000 atmospheres at 3000° C.

Hall's apparatus develops pressure in three dimensions. Each of four anvils that compress the sample has triangular faces. Hydraulic rams drive the anvils together along lines normal to the triangular faces. Using four anvils provides a relatively large work­ing volume, says Hall, compared with using two opposing anvils with circular faces. Heat is supplied to the sample from electrical resistance heaters be­neath the anvil faces. When in opera­tion, each anvil is surrounded by a steel binding ring for added support to op­pose transverse tensile forces.

A pyrophyllite tetrahedron serves a multiple purpose—pressure transmitting medium, thermal and electrical insula­tion, and provides the necessary com­pressible gasket. (Pyrophyllite is a naturally occurring hydrous aluminum silicate, often called Tennessee Grade

A Lava. Since it is readily machin­able, it is an excellent material to use here, says Hall.) The sample con­tainer, a tube which doubles as an elec­trical resistance heater, rests within the pyrophyllite tetrahedron.

Thermocouples inserted through holes in the sample indicate tempera­ture. The friction of the pyrophyllite on the fine thermocouple lead wires is enough to hold them in place during high pressure operation.

In opération, the pyrophyllite tetra­hedron centers on the anvil faces, which are painted with rouge to in­crease friction and prevent extrusion of pyrophyllite. Since the triangular faces of the pyrophyllite tetrahedron are larger than the triangular anvil faces, some pyrophyllite is forced between the sides sloping from the anvil face, auto­matically forming a gasket. Continued motion of the anvils compresses· this gasket first, then the tetrahedron, and finally the sample.

The apparatus is calibrated by ob­serving abrupt changes in electrical re­sistance of materials such as bismuth as pressure increases. Since the vol­ume of the pyrophyllite tetrahedron is about 16 times as large as the sample container, differences in compressibility of various liquid and solid samples do not appreciably affect the pressure cali­bration.

• Other W a y s Possible. Hall also points out some other possible ways to obtain very high pressures. One is based on the idea of "pressure genera­tion from within," A simple piston and cylinder arrangement could be clamped shut with no portion of the piston pro­truding from the cylinder—eliminating the piston breaking problem—and could withstand 500,000 atmospheres, if the pressure could be somehow generated from within, says Hall.

Materials that expand upon freezing would work. One end of a sample might be frozen to provide expansion with its own thermal insulation per­mitting heating of the other end under pressure conditions. Bismuth gives a maximum theoretical pressure near 18,000 atmospheres and actually has been used in some places. But ger­manium offers the best prospects for obtaining very high pressures this way. By freezing germanium, theoretical pressures could go as h igh as 180,000 atmospheres, predicts Hal l .

Chemical reactions, phase changes, and expansion due to heating also offer possibilities for obtaining high pressure within a confined volume. •

M A Y 5, 1958 C & E N 4 1