Using fuzzy sets for data interpretation in natural analogue studies

Conference · Tue Jul 01 04:00:00 EDT 2008

OSTI ID:21326169

De Lemos, F L ^[1]; Sullivan, T ^[2]; Hellmuth, K H ^[3]

---

1. CNEN- National Nuclear Energy Commission (Brazil)
2. BNL-Brookhaven National Laboratory (United States)
3. STUK- Radiation and Nuclear Safety Authority (Finland)

Natural analogue studies can play a key role in deep geological radioactive disposal systems safety assessment. These studies can help develop a better understanding of complex natural processes and, therefore, provide valuable means of confidence building in the safety assessment. In evaluation of natural analogues, there are, however, several sources of uncertainties that stem from factors such as complexity; lack of data; and ignorance. Often, analysts have to simplify the mathematical models in order to cope with the various sources of complexity and this ads uncertainty to the model results. The uncertainties reflected in model predictions must be addressed to understand their impact on safety assessment and therefore, the utility of natural analogues. Fuzzy sets can be used to represent the information regarding the natural processes and their mutual connections. With this methodology we are able to quantify and propagate the epistemic uncertainties in both processes and, thereby, assign degrees of truth to the similarities between them. An example calculation with literature data is provided. In conclusion: Fuzzy sets are an effective way of quantifying semi-quantitative information such as natural analogues data. Epistemic uncertainty that stems from complexity and lack of knowledge regarding natural processes are represented by the degrees of membership. It also facilitates the propagation of this uncertainty throughout the performance assessment by the extension principle. This principle allows calculation with fuzzy numbers, where fuzzy input results in fuzzy output. This may be one of the main applications of fuzzy sets theory to radioactive waste disposal facility performance assessment. Through the translation of natural data into fuzzy numbers, the effect of parameters in important processes in one site can be quantified and compared to processes in other sites with different conditions. The approach presented in this paper can be extended to facilitate a comparison between sites. For example, the degrees of membership can be a measure of similarities between sites and, consequently, confidence in validation tests of models can be enhanced. (authors)

• Research Organization: WM Symposia, 1628 E. Southern Avenue, Suite 9 - 332, Tempe, AZ 85282 (United States)
• OSTI ID: 21326169
• Report Number(s): INIS-US--10-WM-08451
• Country of Publication: United States
• Language: English

Related Subjects

58 GEOSCIENCES
FUZZY LOGIC
MATHEMATICAL MODELS
NATURAL ANALOGUE
RADIOACTIVE WASTE DISPOSAL
RISK ASSESSMENT