A Generic GoldSim Performance Assessment
Note: Clicking on the illustrations will bring up larger versions.
This model is presented as an example of how environmental modeling can be done using the GoldSim platform. This particular case is a radiological performance assessment (PA), as would be done in support of a disposal facility for low-level radioactive waste (LLW). Commercially operated LLW disposal facilities are licensed by the U.S. Nuclear Regulatory Commission (or agreement states). The U.S. Department of Energy operates such facilities on its own behalf as well. Both entities require a PA be done in order to assess the potential future risk to human receptors.
The model includes many typical features and processes that would be part of a PA, but it is entirely fictitious. This does not represent any particular site and is meant to be a generic example. A practitioner could, however, start with this model and, by adding site-specific features and parameter values (distributions), use this model as a starting point for a real model to be used in real decision making.
This method of constructing PAs offers several advantages to the traditional technique, which involves constructing and running many diverse computer programs, which generally do not communicate directly with each other, to run a series of calculations for each part of the model. This common method is not transparent, does not allow for feedback between the different modeled parts, and is difficult to present to the decision maker and the public. In general, past PAs have been solely deterministic in nature as well, meaning that they produce a single result (e.g., dose to a receptor) which does not capture the uncertainty inherent in the process.
New software like GoldSim, combined with ever-increasing computer power, brings the art of Performance Assessment to new levels of sophistication, transparency, and utility. Regulators and other decision makers, as well as the general public and other stakeholders, will benefit from the improvements offered by this method of modeling. Perhaps the most significant innovation is the integration of the state of knowledge about a site (including its radiological inventory and environmental processes at work) into the model itself. The uncertainties inherent in the analysis are reflected in the range of results obtained by the model.
The Conceptual Site Model
This generic application assumes a fairly simple site in order to present ideas. Much more complex sites can be (and have been) modeled by extending the basic techniques introduced here. The present conceptual site model (CSM) assumes a simple trench or pit into which LLW has been disposed. No credit is given to a trench liner or engineered materials in the cover, as such materials are generally either not used or are not expected to last for a significant part of the tens of thousands of years that the disposed materials may present a hazard.
This LLW trench is situated near the ground surface, with a few meters of earthen cover, in a common configuration called shallow land burial (SLB). Water from precipitation and run-on is allowed to infiltrate, mingle with the disposed waste, leaching its radiological constituents, and continue downward through the unsaturated zone to the water table. There the contaminated water mixes with an aquifer that supplies a drinking water well. Biological processes are also modeled, including plant uptake and redistribution of contaminants to the ground surface and animal burrowing, which moves contaminated soils to the surface and back down through the soil column as burrows collapse.
Two human receptors are modeled, a transient occupant and a local resident who engages in light agriculture in contaminated soil. The transient is assumed to spend a limited amount of time walking around the site, exposed to dust and "ground shine" but not local water or foodstuffs. The resident spends much more time on the site, eats locally produced crops and animal products, and drinks water from the aforementioned well. This resident scenario is a standard for most PAs.
The GoldSim Model
The CSM is translated into a mathematical representation, focused on following the fate and transport of the radiological constituents from the waste form to the receptor. This mathematical abstraction is essentially a large "word problem," where one translates a narrative description into parameters and equations. Once the mathematical model is devised, it can be coded into a computer model. In this case, GoldSim is used to document both. Since GoldSim is a "whiteboard"-style program, meaning that it is graphically based using "pages" which can contain text, links to sources of information, and modeling elements all together, the model can be used to document itself. Modeling elements include data inputs (either single-valued or as probability distributions), equations from the mathematical model, and specialized elements to represent discrete volumes of environmental media (e.g., soil or rock) or even the waste form itself. Result elements are used to display results in a variety of formats, usually as time histories of contaminant concentrations or doses to the receptors.
The Generic PA GoldSim model can be run using mean (average) values of the parameters, which will produce an average result, useful for getting the general idea of the future risk posed by the waste disposal facility. The model can also be run probabilistically, using any number of realizations (typically hundreds or perhaps even thousands for a sufficiently complex model), each of which represents a possible outcome, given the uncertainties in the input parameter values, and perhaps in parts of the model itself. (This is called model uncertainty, and it, too, can be implemented using GoldSim.) The probabilistic, or stochastic, results reflect the state of knowledge about the site and its behavior, and therefore a more honest evaluation of the potential future risk it poses. In this case, the resulting doses have a spread of about four orders of magnitude.
These results can be used for sensitivity analysis, which can identify the uncertain parameters that are most significant in determining the dose, for example. A decision maker can then evaluate whether it makes sense to devote further resources to reducing uncertainties, or if the uncertainty reflected by the model is acceptable for the purposes of decision making (e.g., deciding whether the site requires further engineering or controls in order to be protective of the public).
Model and Software Downloads
The full model requires the use of the commercial version of GoldSim to run or view it. Unfortunately, the Academic version is limited to 500 modeling elements, and I could not make the model that small. If you do not have a license for running GoldSim, you can still try the full version for 30 days using the GoldSim Evaluation Version. This will allow you to fully manipulate the model and edit it as well.
A log of version upgrades:
- v1.0: The original release
- v1.001: Modified to be fully metric (using Bq and Sv instead of Ci and REM)
- v1.1: Updated to better show sensitivities and to run under GoldSim v9.1
- v1.2: Modified to run under GoldSim v9.20
- v1.3: Modified to run under GoldSim v9.60
- v1.31: Modified to run under GoldSim v10.02; several bug fixes
- v1.4: Modified to run under GoldSim v10.50; minor improvements
- v2.0: Modified to run under GoldSim 11.1
- v2.1: Modified to run under GoldSim 12.0
Here are links to the full model and software needed to run it:
For those who cannot or do not desire to use GoldSim or the Evaluation version, I have made this model into a GoldSim Player-enabled model, so that anyone can run it using the (free) GoldSim Player. The Player version does have its limitations, however, and using the full version of GoldSim Pro will be much more satisfying.
Here is a link to the Player software needed to run it:
It is my hope that this Generic PA model will help to communicate the advantages of probabilistic modeling in a decision making context, and that it will serve as a template for further work.
- John Tauxe