Latest update: May 2, 1997
Analyzing waste storage tank data
from the Hanford Reservation
During phase I of the Hanford waste characterization plan two tanks were
sampled. We will focus on one of these, U-110,
for which we will look at data for just a few of the
analytes sampled, and some of the radiation
measurements.
In order to estimate the mean composition of waste
in tank U-110 based on these data, we have to make two assumptions:
- that the lab can homogenize and sample individual segments;
- that the lab can combine and rehomogenize samples to form the
core composite sample.
As long as these two assumptions are valid,
the composite sample will be representative of the entire core. To
test this assumption, a simulated core composite was formed by
combining data obtained from the individual segments within a core.
The results were statistically compared with the corresponding data
from the core composite sample formed in the lab, and the general
conclusion was that the two data sets agreed well enough to validate
the assumption.
The task is to use these data in order to assess the assumptions and sample
size calculations provided in part I of the case. Among things to consider
are:
- How do you deal with numbers that are smaller than the
detection limit for the analytical procedure?
- How do you deal with incomplete core recovery, for example when there
is hard waste that the sampler cannot recover?
- If different analytes yield different sample sizes, what do you do?
Copyright Notice
This Web page may be reprinted or adapted for academic nonprofit purposes,
providing the sources are accurately quoted and duly credited.
STAT 498: Copyright© 1996-7, Peter Guttorp.
Email corrections or comments to Peter Guttorp.
Last Updated May 2, 1997