Edward O'Donnell
U.S. Nuclear Regulatory Commission
Washington, D.C. 20555

Peter Godwin
University of Maryland
Agricultural Experiment Station
Beltsville, MD

Robert K. Schulz
University of California
Berkeley, California 94720

This study's objective was to assess means for controlling water infiltration through waste disposal unit covers in humid regions. Experimental work was carried out in large-scale lysimeters 21.34 m x 13.72 m x 3.05 m (70 ft x 45 ft x 10 ft) at Beltsville, Maryland. Results of the assessment are applicable to disposal of radioactive waste, uranium mill tailings, hazardous waste, and sanitary landfills.

Three kinds of waste disposal unit covers or barriers to water infiltration were investigated: (1) resistive layer barrier, (2) conductive layer barrier, and (3) bioengineering management. Performance monitoring of cover performance began in 1987 with completion of two bioengineering management covers and monitoring continues to date (1998). Monitoring of the resistive barriers began in 1988 and continued to 1996 and monitoring of the conductive layer barrier extended from 1989 to 1996.

The resistive layer barrier consisted of compacted earthen material (e.g., clay). The conductive layer barrier consisted of a conductive layer in conjunction with a capillary break. As long as unsaturated flow conditions are maintained, the conductive layer will wick water around the capillary break. Below-grade layered covers such as (1) and (2) will fail if there is appreciable subsidence of the cover, and remedial action for this kind of failure will be difficult. A surface cover, called bioengineering management, is meant to overcome this problem. The bioengineering management surface barrier is easily repairable if damaged by subsidence; therefore, it could be the system of choice under active subsidence conditions. The bioengineering management procedure also has been shown to be effective in dewatering saturated trenches and could be used for remedial action efforts. After cessation of subsidence, that procedure could be replaced by a resistive layer barrier or, perhaps even better, by a resistive layer barrier/conductive layer barrier system. The latter system would then give long-term effective protection against water entry into waste without institutional care.

As mentioned in the preceding paragraph, a bioengineering management cover might well be the cover of choice during the active subsidence phase of a waste disposal unit. Some maintenance is required during that period. Final closure, using geological materials, could follow cessation of subsidence. No further significant maintenance would then be required. If the geological material used is solely a clay barrier to water infiltration, the cover will be "sensitive" to imperfect construction or degradation by penetrating roots. The roots will die and decay, causing markedly increased permeability of the clay with the passage of time. A system using a conductive layer under the clay layer as a water-scavenging system will, in comparison, be "robust." Roots will still degrade the clay layer but will not degrade the scavenging layer. A root hole through the conductive layer will be analogous to a hole through a wick. It will do no significant damage. The combination of a resistive layer with a conductive (scavenging) layer underneath is thus less dependent on perfect construction techniques and will be resistant to damage by root invasion. In the absence of subsidence such a system should function effectively for millennia.

Another very useful application of the resistive layer barrier/conductive layer barrier system would be to protect an earth-mounded concrete bunker. In that case, the barrier system would shield the concrete from exposure to flowing water. The resulting stagnant alkaline film of water would tend to protect the concrete from degradation over a long time period. Similarly, a resistive layer barrier/conductive layer barrier system could be used to for mined cavern disposal. If waste were disposed of in fractured rock, this system could be used to divert possible fracture flow water around the waste.

Infiltration of water into waste is the foremost problem associated with near-surface disposal of radioactive waste, uranium mill tailings, and municipal waste. Up to this time, disposal unit covers have generally been constructed from soil materials. In humid areas, these soil or clay covers have generally proved less than satisfactory. Often, the cover itself has served as the principal pathway for water entry into the waste (1). Water infiltrating to buried waste, contacting the waste, then exiting the area can reasonably be expected to be the most important of radionuclide transport agents. Some radionuclides, such as tritium (present as tritiated water), and those present in anionic form or neutral complexes, will essentially move with the flow of water. Others, present as cations, will move much more slowly, but all will move to a greater or lesser degree. Clearly then, it is advantageous to reduce water infiltration to buried waste to as low a level as reasonably achievable. It was the purpose of our work to examine and demonstrate various approaches for achieving that goal.

Three kinds of waste disposal unit covers or barriers were investigated:

1. Resistive Layer Barrier -- the resistive barrier is one that resists infiltration and may consist of compacted earthen material such as clay, a geomembrane, or some combination of them.
2. Conductive Layer Barrier -- The conductive layer barrier consists of a porous medium (the conductive layer) underlain by a capillary break. The conductive layer must be able to conduct reasonable amounts of water under a negative matric potential, that is, under unsaturated flow conditions. Suitable materials might include a fine sandy loam or a fine grained sand layer placed over a capillary break.
3. Bioengineering Management -- This cover differs from the other covers in that it is a surface cover consisting of impermeable panels which are placed to promote run-off and to limit infiltration. Drought resistant vegetation is placed between the panels to scavenge moisture below the panels.

Resistive layer barriers and conductive layer barriers are vulnerable to subsidence and they must fail if subjected to substantial shearing caused by waste degradation. Re-establishment of a layered system after subsidence failure is a difficult undertaking and is exacerbated by the increasing complexity of the layered system. The failure potential of in-ground layered systems during the subsidence period argues for development of an easily repairable surface barrier for use during that period. To that end, a procedure called "bioengineering management" was developed (1). The bioengineering management technique utilizes a combination of engineered enhanced run-off and moisture-stressed vegetation growing in an overdraft condition to control deep water percolation through disposal unit covers. An artist's conceptual drawing is shown in Fig. 3, referenced in (2).

Performance of all three types of barriers continued up to September 1996 when the initial period of funding ended and the results are reported in (1). The site was turned over to the Department of Interior in September 1996 for other types of experiments. However monitoring was reestablished on the bioengineering covers in April 1997 and this will be the focus of this report.

In this section we will discuss experiments that have been conducted in large-scale lysimeters at a humid region site in Beltsville, Maryland. The placement of the lysimeters is shown in Fig. 4, reference (5).

In bioengineering management the necessary run-off is provided by features installed at or above the soil surface rather than within the profile. The procedure, described by Schulz et al. (1), was designated bioengineering management. Its principal advantage is that subsidence can easily be managed by relatively simple, inexpensive maintenance of the aboveground features rather than by difficult reconstruction of below-ground layers. It should be noted that, after a length of time sufficient so that the organics have decayed and the waste containers have completely failed, subsidence will cease and a layered system could be then installed which could last over geological time periods.

In essence, the bioengineering management technique utilizes a combination of engineered, enhanced run-off and stressed vegetation in an overdraft condition to control deep water percolation through disposal unit covers. To describe it further: if a waste burial site is selected so that incoming subsurface flow is negligible, then precipitation is the sole source of input water. In a simplified model, that water has three possible fates: [1] evapotranspiration, [2] run-off, and [3] deep percolation. Evapotranspiration has a definite limit, governed by energy input. Ideally, deep percolation should be zero, leaving only the run-off component available for unlimited manipulation. Positive control of run-off becomes difficult with the use of compacted porous media trench caps as the sole barrier to water infiltration. The compacted material tends to become more permeable with the passage of time, due to fractures caused by waste subsidence and from the inexorable process of root growth, followed by death and decay of the roots, thus creating water channels. Evapotranspiration cannot then use all of the infiltrating water, and water percolates downward to the waste. As stated before, evapotranspiration has a theoretical maximum dictated by solar energy input to the system; only run-off remains available for nearly unlimited management. This run-off can be surface or subsurface, as long as it occurs before water reaches the waste.

Surface run-off can be managed to as high as 100% by means of a perfect, leak-proof roof, which is expensive and hard to guarantee. Alternatively, adequate but not total run-off can be engineered rather inexpensively by using an impermeable ground cover over part of the surface to achieve high and controlled levels of run-off. Vegetation planted between areas of impermeable cover will extend over the cover to intercept incoming solar energy to evaporate water. Roots will extend under the cover in all directions to obtain water.

Such a system can be visualized by imagining a supermarket parking lot, where trees are planted in islands, surrounded by concrete curbs, within an extensive paved area. In this case, the trees are maintained in a drought environment due to the small soil surface available for infiltration of precipitation. The paving, along with the curbing around the trees, causes run-off of most precipitation. Aboveground, the tree's branches and leaves extend over the parking lot and intercept incident solar energy. Beneath the surface, the roots, in a drought state, explore outward under the paving for any available water. Utilizing this concept, it should be possible, by combining engineered run-off with vegetation, to maintain the soil profile in a potential overdraft condition on a yearly basis.

Initial investigations of the bioengineering management technique were carried out in lysimeters at Maxey Flats, Kentucky. Results obtained in seasonal 1984-1985 and 1985-1986 were reported by O'Donnell et al. (3). In that work, a fescue grass crop was used with an engineered cover of stainless steel. Following seasonal 1985-1986 the grass cover was removed, a new stainless steel engineered cover was constructed, and Pfitzer junipers were planted in the lysimeters. After the junipers were established, percolation data were again collected in 1988 and reported by Schulz et al. (7). The woody junipers were excellent in preventing deep percolation of water in the lysimeter.

The encouraging initial results obtained in the Maxey Flats lysimeter experiment led to the establishment of a large-scale field demonstration at Beltsville, Maryland in Fig.4, referenced in (5). Figure 5, referenced in (5), is a photograph of lysimeter 1, bioengineering management, taken in September 1996, ten years after planting of the Pfitzer junipers. Alternating panels of aluminum and fiberglass were used as the hard cover. These plots, or lysimeters, are 21.3 m (70 ft) long by 12.7 m (45 ft) wide, and the bottoms are 3.05 m (10 ft) below grade. Figure 6, referenced in (5), shows a side view of construction details of lysimeters 1 and 2 (bioengineering management). The only difference between the two was the initial water level in the lysimeters. The water level was 90 cm above the bottom of lysimeter 1 and 190 cm above the bottom of lysimeter 2. The water level in the lysimeters simulates the water table in a flooded disposal cell. In addition to the two bioengineered lysimeters, two reference lysimeters (3 and 4) were initially constructed. They were similar to the former, except that they were merely planted with fescue grass. No hard cover was present, but surface slopes were similar to the two bioengineered lysimeters (i.e., a slope of 1:5). Performance data for the reference lysimeters are given in Fig. 7, referenced in (5).

The water level in the two reference plots or trenches (lysimeters 3 and 4) rose until it was near the surface. At that time, water was pumped from the lysimeters to keep them from running over. The graphs of the water tables (i.e., water levels) in the bioengineered plots (lysimeters 1 and 2) show an entirely different story, as evidenced in Fig. 8, referenced in (5). In both cases, the water table was eliminated. It appears that the bioengineering approach could prevent water infiltration to a disposal unit. It also could be used for a remedial action in dewatering existing problem sites such as Maxey Flats.

On February 4, 1988, lysimeter 4 was pumped out to prevent overflow. It was then discontinued as a reference lysimeter and converted to a rock-surfaced, resistive-layer barrier plot. Lysimeters 1 and 2 (bioengineered) and lysimeter 3 were continued. A summary of run-off, evapotranspiration, and pumping from those three lysimeters is given in Fig. 1.

Figure 1 shows that there was very little run-off from the grass-covered plot. Most of the precipitation was disposed of, via evapotranspiration, by the fescue crop, but this was not adequate to prevent the rise of the water table. Table I gives the run-off, evapotranspiration, and deep percolation in the bioengineered plots during the past ten years. There was no deep percolation during this period. Until seasonal 1993-1994 the evapotranspiration had been rising annually, probably as a result of the greater vegetative canopy intercepting a greater percentage of the precipitation. In 1988, 1989, 1990, 1991, 1992, 1993, 1994, 1995, 1996, and 1997 the run-off percentages were 80, 74, 70, 67, 63, 61, 61, 58, 57, and 55% respectively. During 1989, the water table was completely eliminated in both plots. This is show as a graph in Fig. 8, reference (2).

In addition to rainfall, run-off, and evapotranspiration measurements discussed above, neutron-probe soil-moisture measurements were made continuously to monitor soil moisture changes in lysimeters 1, 2, 4, 5 and 6. The neutron probe measurements will indicate whether there is a gain or loss of moisture from the soil profile or, perhaps, steady-state situation, where there is little or no net gain or loss of soil moisture during a year. A steady-state situation with relatively constant-moisture "dry" soil above waste would be highly desirable with a bioengineered cover. There would then be a large safety margin to protect the waste from infiltrating water.

Neutron probe apparatus, as supplied by the manufacturer, is calibrated against moisture measurements in sand. Such calibration is of unknown accuracy when applied to soil measurements. For this reason, the probe was calibrated using the same soil as in the lysimeters. Six hundred and twenty-eight kilograms (1400 lbs) of soil were placed in a weighing lysimeter, and measurements were made over a seven year period. Calibration data obtained using the weighing lysimeter are given in Table II, reference (2). The resulting curves, depicting the factory calibration and the weighing lysimeter calibration, are given in Fig. 10, reference (2). It is evident that use of the factory calibration on sand would result in a very large error in soil moisture determination.

Results of some neutron probe measurements are shown in Fig. 2 for bioengineered lysimeters 1 and 2. The data are plotted as volumetric moisture content, as a function of soil depth, on specific dates. Eleven widely spaced measurement dates are shown, for clarity. From inspection of the figure it is seen that, at the start of the experiment in July, 1987, the moisture content of the soil increased with depth until the water table was reached, then became constant. By July, 1989, the water table had been eliminated from both lysimeters, and the soil profiles were drying out. However, the soil moisture content, although much lower in the soil profile than in July, 1987, still increased with depth. This same relationship was still evident in August, 1996, although the soil profile had become still drier.

Figure 12, referenced in (2), shows the moisture content of the soil profiles in lysimeters 1 and 2 at the end of each seasonal year. Following the complete removal of the water tables during the 1987-1989 period, the soil profiles were dried out further during the ensuing years. However, an unanticipated result turned up in lysimeter 1 at the end of seasonal 1993-1994. The moisture content of the soil profile increased slightly. To shed light on that result, the moisture content in the soil profiles at four depths were plotted monthly along with monthly rainfall data (Fig. 3). Here we see seasonal cyclical variations in moisture content in the soil profiles, with peak moisture concentrations occurring in the early spring, following periods of significant rainfall and minimal evapotranspiration. That cycling is both obvious and expected. What was totally unanticipated was the increase in the moisture peaks in lysimeter 1 in each of the 3 seasonal years leading to 1994-1995. The increasing amplitudes of the moisture curves did not appear to be a result of rainfall variations, nor were they present in lysimeter 2. The aforementioned trend did not continue during the past three years, seasonal 1994-1995, seasonal 1995-1996, and seasonal 1996-1997.

The results to date indicate that bioengineered closure, as described in this experiment, would maintain the cover over buried waste in a "dry" steady-state condition. This would not only prevent water from percolating down to the waste, but would do so with a large safety factor.

As previously mentioned, on February 4, 1988, lysimeter 4 was pumped out, discontinued as a reference lysimeter, and converted to a rock-surfaced, resistive-layer barrier plot. The primary reason for constructing that particular cover is the likelihood of such covers being used for uranium mill tailings. An end view of that plot or lysimeter is shown in Fig. 14, referenced in (2). This lysimeter was completed in the fall of 1988, and data collection (measuring performance) was carried out. The most important information to be gained here was the relative weighing of the advantages and disadvantages of rock surface versus. a vegetated surface.

In addition to the UMTRA or rock-surfaced, resistive-layer barrier plot, a vegetated resistive layer barrier plot was constructed. The primary purpose of this plot was for comparative measurements. Essentially, this plot was similar to the rock-surfaced plot except that topsoil replaces the rock layer, and the plot is planted with fescue grass. A diagram of this plot is given in Fig. 15, referenced in (2).

In Fig. 1, the fate of precipitation in the UMTRA and grass-covered, clay-layer lysimeters is given. There was more than twice as much run-off from the rock-covered plot than that from the grass-covered plot. The data show no deep percolation through the clay layers through seasonal 1993-1994 in either lysimeter and there is little indication as to how much safety margin has been offered. In seasonal 1994-1995, 0.10 cm (0.04 in) passed through the UMTRA cover. In seasonal 1995-1996 no water passed through the cover. It is not known how consistently such near-perfect clay barriers would be installed in a routine operation. That remains a problem for future consideration.

Another concern is the possible drying out of clay barriers. If this were to happen, the clay layer would not be as efficient a barrier for preventing radon escape as planned in the UMTRA application. In addition, drying out of the clay layer could lead to cracking, leading to subsequent leakage prior to resealing by wetting. Figure 4 gives the volumetric moisture content of clay in the rock-covered (lysimeter 4) and the grass-covered (lysimeter 6) plots. In no case did the clay layer dry out significantly. On the contrary, in the UMTRA or rock-covered plot, which was devoid of vegetation, there was a slight increase in moisture content with time, suggesting that some leakage of water through the clay layer would occur in the future. A slight leakage did occur during seasonal 1994-1995. Lysimeter 6 has a clay layer and a grass cover. In this case, no increase in moisture content was observed. On the contrary, to date, the moisture content of the clay layer seems to be in a rather steady state, taken over the 7 year period of measurement.

If we consider the case of water flowing downhill in an unsaturated porous medium, we have the case shown in Fig. 17, reference (2). The "holes" shown in the diagram could be a rock layer, affording a capillary break or capillary discontinuity Fig. 18, reference (2). Under appropriate conditions, water everywhere in these cross-sections will be under tension, and there will be no leakage. This might then serve as an excellent means of protecting waste by conducting water around the waste. Figure 18, reference (2), simulates a conducting porous medium, such as a fine sandy loam soil, lying smoothly on top of a rock layer. Problems with water flow under saturated conditions could certainly arise where a less than smooth surface ends up being constructed as depicted in Fig. 19, reference (2). That is, what happens if imperfections are constructed so that "pockets" of soil extend down into the rock layer? Figure 19 represents that case. Again, there will be no leakage, provided conditions are such that the water in all parts of the conductive layer remains under tension.

The big question is, can conditions required to maintain the necessary soil water tension be practically maintained while using this procedure to effectively protect waste disposal units? To answer this question the apparatus schematically depicted in Fig. 20, referenced in (2), was constructed, i.e., a "soil beam." Several miniature soil beams Fig. 21, referenced in (2), were constructed for use in the laboratory so that a variety of candidate conductive-layer materials could be quickly evaluated.

A number of materials were evaluated using the miniature soil beams. It was quickly established that it would be necessary to construct a resistive layer barrier above the conductive layer barrier to have a practical system. The standard was set that the resistive layer barrier have an easily achievable conductivity of not greater than 10^-6 cm/sec. On this basis it was found that material such as fine sandy loam could provide an effective conductive layer barrier, that is, conduct around the waste 100% of water percolating through the resistive layer. However, the measurements showed that such materials would not provide the desired (factor of 10) safety margin.

Further investigations turned up a material, diatomaceous earth, that would fit these requirements. Measurements of tension versus. distance of flow are shown in Fig. 22, referenced in (2).

The results of this experiment in the 137 cm (4.5 ft) long beam suggest that, as long as the flow rate is no greater than 4.2 x ^10-4 cm/sec, the soil water will remain under tension regardless of the soil beam length. These results show that with the use of diatomaceous earth for the conductive layer and following the easily achievable standard set above for the resistive layer, it should be possible to construct a barrier that would allow no water leakage to a waste disposal unit. However, before final selection of the diatomaceous earth as the conductive layer material, we believed it to be prudent to conduct tests in a large-scale soil beam. The large beam, shown in Fig. 23, referenced in (5), has a soil beam length of 6.4 m (21 ft). As shown in Fig. 24, referenced in (5), a matric potential of about -15 to -20 cm of water is maintained over the entire 6.4 m length of the beam when the flow rate does not exceed 3.1 x 10^-4 cm/sec.

The studies carried out in the large soil beam closely confirmed the data obtained in the miniature beam. Accordingly, diatomaceous earth was used as the conductive layer material in the demonstration lysimeter (lysimeter 5). It has been estimated that purchasing and shipping the diatomaceous earth to a job site any place in the United States will add about $0.50 per ft³ of disposed waste. This is over the cost of using locally obtained soil, and based on waste being 3.05 m (10 ft) deep.

After the time-consuming task of selecting the conductive layer material was accomplished, a resistive layer barrier over a conductive layer barrier was constructed in lysimeter 5. It was completed in January, 1990. A local clay from Beltsville, Maryland, the Christiana Clay, was selected as the resistive layer barrier. Testing has shown this material more than meets specifications. A cross-section of the cover system is shown in Fig. 25, reference (2).

Performance of this cover is shown in Figures 1 and 4 (lysimeter 5). Until seasonal 1993-1994 the cover system was 100% effective in preventing water movement downward through the cover. In seasonal 1993-1994, 0.13 cm (0.05 in) of water passed through the cover to the pan shown in Fig. 25, reference (2). In seasonal 1994-1995, 0.21 cm (0.08 in) passed through the cover. Although that amount is an extremely small percentage of the total rainfall, in theory no water should have percolated through the cover to pan. It is possible that the cover system was compromised by the instrumentation installed to measure performance. No water passed through the cover in seasonal 1995-1996. Thus the total leakage was only 0.34 cm (0.13 in) over the 6 year period. That is 0.34 cm (0.13 in) percolation out of a total rainfall of 665 cm (261 in).

For the Beltsville study, diatomaceous earth was selected for the conductive-layer material, based on a combination of performance and cost considerations. Based on these two considerations only, diatomaceous earth would still be the material of choice, particularly since it has a much lower bulk density than sand and is therefore less expensive to ship. However, the engineering properties of sand are better known, thus sand may be more attractive to some installers. Therefore, we have conducted further studies with various sands. Results of studies of the unsaturated flow characteristics of four different sands are given in Fig. 26, reference (2). All these sands exhibit unsaturated flow rates that are about twice that of the diatomaceous earth at any given negative matric potential. The particle size distribution of the four sands is given in Table III, reference (2). The mortar sand, for example, had the narrowest particle size range, and the foundry sand had the widest particle size distribution, although the particle size distribution did not have an important effect on the flow rates reported in Fig. 26 (reference 2). The Nevada dune sand and the Kelso dune sands are from large eolian deposits in the Nevada and California deserts, respectively. The Kelso deposit has been mined commercially.

The three procedures described in the Introduction may be used singularly or in combination to protect disposal units from percolating water. The principles apply equally to above-ground or below-ground disposal. For example, a combination of covers 1. and 2., described in the Introduction, could be ideal for a stabilized, shallow land burial facility, whether it is above or below ground; e.g., the subsurface disposal could be in below-ground vaults, and the above-ground disposal units could be earth-mounded concrete bunkers. A combination of a resistive layer over a conductive layer in a concrete bunker or above-ground application is shown in Fig. 27, reference (5). The resistive (clay) layer is the primary barrier. The small amount of water passing through the clay layer will be diverted around the concrete bunker by the conductive layer. This cover over the concrete bunker can, in theory, be 100% effective, shielding the bunker from exposure to flowing water. This would result in a film of stagnant alkaline water at the gravel/concrete interface. The presence of this high pH,stagnant water would tend to protect the concrete from degradation over a long period.

The bioengineering concept could be advantageous for either a tumulus or shallow land burial unit that would be likely to exhibit subsidence. If desired, and after subsidence has ceased, a combination of covers 1. and 2. could be constructed with geological materials to give extremely long-term isolation without further maintenance. Another possible application of a combination of covers 1. and 2. described in the Introduction is shown in Fig. 28, reference (2). Here, radioactive waste is emplaced in a tunnel excavated in rock. If a fracture were present in the rock, and fracture flow occurred, the combination of a resistive layer and a conductive layer could provide excellent isolation of the waste from flowing water. Figure 29, reference (2), depicts an application where only very low flow rates need be protected against (essentially, dropwise fracture flow). Here, the system could be simplified so that only a conductive layer with a capillary break is necessary.

Funding for this project was provided by the Waste Management Branch, Division of Regulatory Applications, Office of Nuclear Regulatory Research. Dr. Sher Bahadur is Chief of the Waste Management Branch, and Dr. Joseph A. Murphy is Director of the Division of Regulatory Applications.

1. R.K. SCHULZ, R.W. RIDKY, and E. O'DONNELL, "Control of Water Infiltration Into Near Surface LLW Disposal Units-A Discussion", NUREG/CR-4918 Vol. 2: 24 pp., U.S. Nuclear Regulatory Commission, Washington, DC 20555, (1988).
2. R.K. SCHULZ, R.W. RIDKY, and E. O'DONNELL, "Control of Water Infiltration Into Near Surface LLW Disposal Units - - Final Report on Field Experiments at Beltsville, Maryland", NUREG/CR-4918 Vol. 10: 19 pp., U.S. Nuclear Regulatory Commission, Washington, DC 20555 (1997).
3. E. O'DONNELL, R.W. RIDKY, and R.K. SCHULZ, "Control of Water Infiltration Into Near Surface LLW Disposal Units", in Oak Ridge Model Conference, CONF-871075, Vol. 1, Pt. 3: 355-384 pp, NTIS, Springfield, VA (1987).

BACK