Comparison of Empirical Relationships for Hydraulic Conductivity Using Grain Size Distribution, Packing, and Porosity Information from the Boise Hydrogeophysical Research Site, Boise, Idaho

Publication Date


Type of Culminating Activity


Degree Title

Master of Science in Geology

Major Advisor

Warren Barrash


Hydraulic conductivity, or K, is widely accepted to be the most important parameter for understanding groundwater flow and contaminant transport in aquifers. Most subsurface hydrology research employs methods for determining the magnitude and distribution of K using analytical scales that vary from centimeters to meters. Heterogeneity in K significantly affects contaminant transport.

This thesis focuses on the heterogeneity of K in coarse grained alluvial sediments. Several empirical models for the calculation of K, ranging from simple to complex, are used for three of the 18 wells at the Boise Hydrogeophysical Research Site (BHRS), which is a research wellfield located approximately 15 kilometers east of Boise, Idaho. All the models use grain size information, and some of the models incorporate porosity information, to estimate K values of the BHRS aquifer at the core (sub-meter) scale. The results are then compared against each other and against independent K data from the site.

The empirical models' K values span six orders of magnitude, and the weighted averages of K for most relationships result in K estimates larger than the averages from the fully penetrating pumping test analysis and tracer test. The overestimation is attributed to using effective grain sizes that are not appropriate for coarse conglomeratic deposits with a large range of grain sizes. This study selected a K estimation approach that considers the BHRS sediments to be mixtures of interconnected sand-to-fine-pebble, or "matrix" regions, that fit in the spaces between larger cobbles that form the framework of most of the aquifer. K values for each core sample are estimated by computing the effective grain size (d10) and porosity of the matrix material, since flow cannot occur through cobble grains.

Upscaled K values were calculated by taking the weighted average of the individual K values (36 total values per well) from core for a given well. The upscaled K value for each of the three BHRS wells examined in this thesis is compared to the average K value from (a) a fully penetrating pumping test at the same well (0.05, 0.06, and 0.08 cm/s for C1, B2, and B5 respectively), and (b) tracer test results (0.025 cm/s for elevation interval 839.l to 837.6 meters and 0.033 cm/s for elevation interval 837.6 to 837.1 meters above sea level) at well A1. Of the seven models used in this thesis (Hazen, Beyer, United States Bureau of Reclamation, Kozeny-Carman, Terzaghi, Clarke, and Koltermann and Gorelick), the weighted average K for the fractional packing model (Koltermann and Gorelick) comes very close to the average K values for the fully penetrating pumping test results, while the other models are within a factor of 2 - 5 lower for the effective grain sizes used in these models. The closer match of Koltermann and Gorelick to pump test results can be attributed to coming closer to the calculated de value that is likely closer to the appropriate values. It is apparent that d10 is not always the most appropriate effective grain size to use for the calculation of K. Back calculating from the pump test results to find the optimal de grain size indicated that the optimal de grain size varies significantly from one model to the next (i.e., d4 for Hazen in well C1 to d19 for Terzaghi and Kozeny-Carman in well B2).

The modified K methods used here work very well at the BHRS and this approach represents a broadly applicable method of K estimation for coarse grained sediments.

Files over 30MB may be slow to open. For best results, right-click and select "save as..."