TM 5-818-4/AFM 88-5, Chap. 5
tablished. An example of sensitivity error would be the
quire separate analysis for each backfill material type
nuclear density device that is capable of determining
and compaction effort, complete random selection of
densities only to within 3 to 5 pounds per cubic foot of
test locations, and a large number of control tests as
true density. The second type of error relates to
compared with the conventional decision method. In
constant deviations between measured and true densi-
addition, statistical methods include water content
ty. Constant deviation errors can be corrected by cali-
control, which is not normally included in military
brating test equipment against known densities.
specifications.
(2) The theory and details concerning the applica-
(3) Material property errors are primarily limited
to density determinations using either the sand-cone
tion of statistical methods for compaction control are
or the rubber-balloon method in sands. When a soil is
well developed. Figure 7-1 shows a sequential inspec-
physically sampled during the process of conducting
tion plan example of how the end results of a statis-
an in-place density measurement using these two
tical analysis might be used for the purpose of accep-
methods, a shearing action of the soil is unavoidable.
tance or rejection. In this example, it was established
Cohesionless soils are sensitive to volume change
by statistical analysis that adequate densities could
during shear, dense sands tend to expand and increase
probably be obtained with reasonable confidence by a
in volume, and loose sands tend to contract and de-
given compaction effort for desired water contents
crease in volume. Errors of this nature cannot be
ranging from 3 percentage points below to 1 per-
quantified or detected in the field. However, such
centage point above optimum. It was also established
errors can be as high as 6 percent for sand using the
that a density corresponding to 95 percent of CE 55
rubber-balloon method for volume measurements.
maximum dry density was the minimum acceptable
density based on required engineering performance of
h. Acceptance or rejection. The inspection person-
the backfill. The sequential inspection plan consists of
nel have the responsibility to accept or reject the back-
examining, in sequence, single tests that are obtained
fill or any part thereof based on the quality acceptance
at random from a segment of the backfill being consid-
control tests. On the surface, this task seems straight-
ered for acceptance or rejection and, for each test,
forward. If a segment of the backfill tested at several
making one of three possible decisions: the segment is
locations for acceptance passes or fails to pass mini-
acceptable; the segment is unacceptable; and the evi-
mum requirements by a wide margin, then it is gener-
dence is not sufficient for either decision without too
ally safe to assume that the backfill within that seg-
great a risk of error as indicated by the retest block in
ment either has or has not been adequately compacted
figure 7-1(a). The reject areas in figure 7-1 indicate
and the acceptance or rejection of that segment can be
conditions that cannot be corrected by additional roll-
made based on the test results. On the other hand, if
ing. The material must be replaced in thinner lifts and
the tests indicated insufficient compaction, the size of
be within the desired water content range before ade-
the affected area may be questionable; it is possible
quate compaction can be achieved with the compaction
that the test(s) represents only a small area and the lift
equipment being used. If the retest decision is reached,
being tested may be sufficiently compacted elsewhere.
an additional test is made at a second random location,
In view of the possible errors associated with control
and the same three decisions are reconsidered in light
tests, tests that indicate marginal passage or failure
of this additional information. If the second test falls
should be treated with caution. The borderline case re-
below the accept blocks, the segment of backfill repre-
quires a close look at several factors: how the result
sentative of that test should be rejected; or if compac-
compares with all previous results on the job, how
tion procedures that have produced acceptable tests in
much compaction effort was used and did it differ
the past have not been altered, then the compaction
from previous efforts, how does this particular mate-
characteristics of that part of the backfill should be re-
rial compare with previously compacted materials, the
evaluated.
importance of the lift location in relation to the entire
(3) The primary advantage of statistical methods
structure, and the importance of obtaining the correct
is that they offer a means of systematically evaluating
density or water content from the designer's stand-
acceptance or rejection decisions rather than leaving
point. When all factors have been considered, a deci-
such decisions entirely to the judgment of the inspec-
sion is made as to which corrective measures are re-
tion personnel. However, if experienced and well
quired. What makes such decisions so difficult is that
trained inspection personnel are available, this ap-
they must be made immediately; time will not permit
proach may not be necessary.
the problem to be pondered. Discussion with design
engineers prior to beginning compaction operations
i. Construction reports. A record should be main-
may help in the evaluation of many of these factors.
tained of construction operations. It is valuable in the
(1) On jobs requiring large volumes of backfill, it
event repairs or modifications of the structure are re-
may be advantageous to base the decision to accept or
quired at a later time. A record is necessary in the
reject on statistical methods. Statistical methods re-
event claims are made either by the contractor or the
7-7
Previous Page
