The decision to bury high level nuclear wastes in deep geological formations
led to the study of the Hanford Nuclear Reservation as one of three possible sites for
the first nuclear waste repository in the United States. To adequately evaluate the
environmental impact of siting nuclear waste repositories in basalt aquicludes, it is
essential to know the effects on parameter identification algorithms of thermal
gradients that exist in these basaltic aquicludes. Temperatures of approximately 60° C
and pressures of approximately 150 atms can be expected at potential repository sites
located at depths of approximately 1000m.
The phenomenon of over-recovery has been observed in some pumping tests
conducted at the Hanford Nuclear Reservation. This over-recovery phenomenon may
possibly be due to variations in the fluid density caused by thermal gradients. To
asses the potential effects of these thermal gradients on indirect parameter
identification algorithms, a systematic scaling of the governing field equations is
required in order to obtain dimensionless equations based on the principle of
similarity. The constitutive relationships for the specific weight of the fluid and for
the porosity of the aquiclude are assumed to be exponentially dependent on the
pressure gradient. The dynamic pressure is converted to the piezometric head and the
flow equation for the piezometric head is then scaled in radial coordinates. Order-ofmagnitude
estimates are made for all variables in unsteady flow for a typical well test
in a basaltic aquiclude. Retaining all nonlinear terms, the parametric dependency of
the flow equation on the classical dimensionless thermal and hydraulic parameters is
demonstrated. These classical parameters include the Batchelor, Fourier, Froude ,
Grashof, and Reynolds Numbers associated with thermal flows. The flow equation is
linearized from order-of-magnitude estimates based on these classical parameters for
application in the parameter identification algorithm.
Two numerical solutions are presented which predict hydraulic head given a
continuous set of flow parameters. The first solution uses a totally numerical finite
difference scheme while the second combines an analytical solution with a numerical
solution. A radial coordinate system is utilized for describing an anisotropic confined
aquifer.
The classical inverse parameter identification problem is solved using an
indirect method. This method is based on the minimization of a objective function or
error criterion consisting of three parts: 1) least-squares error of head residuals; 2)
prior information of flow parameters; and 3) regularization. An adjoint equation is
incorporated into the method to eliminate the need to differentiate the heads with
respect to the parameters being identified, increasing the stability of the algorithm.
Verification of the parameter identification algorithm utilizes both "synthetic",
computed generated input data and field data from a well test for a confined aquifer
within the Columbia Plateau near Stanfield, Oregon. The method used is found to
give parameter estimates which are both stable and unique. / Graduation date: 1992
| Identifer | oai:union.ndltd.org:ORGSU/oai:ir.library.oregonstate.edu:1957/36677 |
| Date | 10 March 1992 |
| Creators | Roley, Kenneth L. |
| Contributors | Hudspeth, Robert T. |
| Source Sets | Oregon State University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis/Dissertation |
Page generated in 0.0016 seconds