Introduction to the API LNAPL Parameters Database

The following discussion provides a brief introduction to the API LNAPL Parameters Database. Additional discussion is provided in the User’s Guide for the database, available from the referenced API website (www.api.org/lnapl).

The API LNAPL Parameters database is a collection of information about samples with measured capillary parameters, as well as other physical parameters. The capillary data availability is key for inclusion in the database since these data are measured far less often than other petrophysical properties. Furthermore, capillary properties are sensitive within the LDRM and other multiphase calculations. The data come from a variety of sites, from service stations to refineries, all having related LNAPL issues requiring data and analyses. The data that have been included in this database were not collected for the purpose of populating a database. They were collected by site technical teams to answer specific questions and to aid in the mitigation and remediation of site-specific problems. Given these facts, it is not surprising that the range of tests performed on each sample and the testing procedures used during the tests differ significantly. Nevertheless, this is currently the most complete set of laboratory measurements of samples

whose properties have been analyzed for the purpose of understanding LNAPL remediation in near-surface aquifers.

The API LNAPL database contains the following types of information:

• Sample capillary parameters for the van Genuchten and Brooks-Corey capillary functions, and the raw data from which they were derived. This information is available for nearly all samples in the database.

• Petrophysical data including density (bulk/grain), porosity, permeability, conductivity, water and hydrocarbon saturations, as available.

• Raw grain size distribution data (weight fraction vs. grain size)

• Grain size at various percentages of the cumulative sample weight (i.e. the grain size at the 10th, 50th, and 90th percentiles).

• Grain size distribution statistics (mean, median, standard deviation).

• Fraction of the sample in various grain size classifications (%sand, %silt, etc.). Typically, if the full raw grain size distribution is available, then all of the summary parameters are also available. For some samples, only summary data are available, while others have no data available.

• Fluid properties (viscosity, density, interfacial tensions). Typically, fluid properties samples are not taken from the same samples where rock properties are measured. Fluid properties samples are provided as a form of site characterization, rather than for direct comparison with other samples within the site.

In many cases, the process of measuring one set of sample properties (grain size) makes the sample useless for measuring other properties (porosity). In general, the collection of measurements attributed to a single sample are collected from slices of a single core not separated by more than 6 vertical inches from the slice where the sample capillarity was determined.

The first step in use is opening the database with Microsoft Access 2000tm_{. Next, one would}

typically go to the “Forms” menu, and select “Query Forms.” From this screen, one can explore the multiphase parameters associated with samples based on the category options shown. The first 3 or 4 general soil or parameter range queries are the most useful for initiating most parameter searches. The remainder become sample-specific, which is useful after an initial survey of parameter ranges has been made.

The API LNAPL Parameters database is used to develop a range of parameter values when one or more soil properties have not been measured. As is the case where measured parameters are available, a range of parameters should be used to bracket possible recovery outcomes using the LDRM. The parameter ranges should then be refined by tuning the model to LNAPL saturation, transmissivity, and recovery data from field (as available). Where field LNAPL data are not available, a bandwidth of results encompassing the full range of measured parameters should be considered and results from specific runs should not be held with a high degree of certainty. For parameters that are absent in the site data, the first step is to identify other available site data that are in one or more ways related to the missing parameters in question. For instance, grain- size distribution may be correlated to permeability and capillarity, and grain-size data can be

used to query the API LNAPL Parameters database for other parameters. More specific queries are desirable, where possible. For instance, if soil parameters are needed and grain-size distribution data are available, one could query the parameters database by those specific grain- size characteristics. Similarly, hydraulic conductivity could be another query factor that has direct hydrogeologic applicability. The least specific query is through soil type description for sites having only boring logs with USCS, USDA, Folk or other common soil designations. While least specific, these are sometimes the only available site data to start from. Since the USCS, USDA, and other soil description systems were developed primarily for purposes other than environmental hydrogeologic applications, these soil designations are not always well correlated to the hydrogeologic parameters of interest.

A high level of detail is not always a requirement because final parameter tuning will depend on comparison of the LDRM predictions to site LNAPL data (recovery, saturation, transmissivity, etc.). Therefore, one can perhaps expect that a detailed evaluation will provide a better starting point for parameterization, but any method of querying the database should eventually be tuned to site observations. In other words, the variations in methods to select the first range of selected inputs are really just fine points that may not in the end make any difference to the final parameter set selected. In any case, whether the database query is highly refined or not, the steps of testing the resultant outcomes against site data will help to narrow potential ranges of parameter input. Application of the database is exemplified in example problem #3 (Section 3.6.3).