Research report summaries 2010–2011

This Web page has been archived on the Web.

Archived Content

Information identified as archived on the Web is for reference, research or recordkeeping purposes. It has not been altered or updated after the date of archiving. Web pages that are archived on the Web are not subject to the Government of Canada Web Standards. As per the Communications Policy of the Government of Canada, you can request alternate formats on the Contact Us page.

Contractors' reports are only available in the language in which they are submitted to the Canadian Nuclear Safety Commission (CNSC).

RSP-0261 – Technical review cost-benefit analysis: Fish impingement and entrainment reduction: Pickering Nuclear Generating Station

The CNSC has raised concerns with cooling water intake structures (CWIS) fish mortality impacts associated with the Pickering Nuclear Generation Station (PNGS). In response to these concerns, Ontario Power Generation (OPG) contracted a cost-benefit analysis (CBA) of CWIS alternatives for the PNGS. A draft of this CBA was completed in December 2009. Subsequently, the CNSC contracted DSS Management Consultants Inc. (DSS) to undertake a technical review of this draft report. The DSS review is limited to the matters set out in the Terms of Reference in Appendix A.

The CBA methodology was developed by economists to assist decision makers in making prudent decisions. CBA is based on the economic principle of efficiency which strives to achieve desired outcomes for the least cost and the maximum benefit. Acceptable levels of fish mortality impacts (entrainment and impingement) for the PNGS CWIS project being considered were already set; as a result, technically, the OPG analysis is a cost effectiveness analysis (CEA) and not a CBA. Nonetheless, the requirements for CEA are the same as those for CBA except that the decision criterion is the least cost to meet pre-decided performance level.

The practice of CBA is well established. Specific guidance for conducting CBAs has been produced by various agencies including the Treasury Board of Canada. As well, the US EPA has produced specific guidance for the conduct of CBAs associated with CWIS. These documents as well as several others have been reviewed as part of this analysis. In particular, a series of specific requirements of CBA that are essential for its proper application are set out (see section 2). These requirements have been used to structure the review of the OPG CBA.

Following is a summary of the deficiencies identified with the OPG CBA. Further details are provided in section 4; cross-references to the sections discussing each deficiency in detail are included.

  • Baseline Conditions (section 4.1) – The assumed outage time for maintenance was not adequately documented to assure its reasonableness.
  • Costing of Technology Alternatives (section 4.3.1) – No breakdown of the capital and labour costs including the quantities and unit costs is provided. No supporting calculations are provided. For this reason, the OPG cost estimates in Attachment B are deemed not to be transparent or replicable.
  • Impingement and Entrainment Rates (section 4.4) – Considerable uncertainty is associated with these rates. Central values and probability ranges should have been estimated and incorporated as part of a Monte Carlo analysis
  • Missing Authorities (section 4.5.1) – Appropriate citations are missing to support the values used for some key parameters. Missing citations for key parameters introduces unnecessary doubt.
  • Missing Rationalisations (section 4.5.2) – Where professional judgement is used two essential requirements must be met; namely, each time professional judgement is used, it must be clearly noted in the documentation. The second is that the rationale for the parameter value derived using professional judgement must be clearly set out. Several cases of missing or inadequate rationalisation were noted.
  • Risk and Uncertainty Analysis (section 4.5.3) – Risk and uncertainty are common in CBA. Good CBA practice demand a systematic approach to the treatment of risk and uncertainty. This essential requirement has not been satisfied by the OPG CBA. The use of conservative assumptions that tend to overestimate environmental benefits partially offsets this deficiency but does not provide adequate understanding of the upside and downside risks.
  • Non-use Benefits (section 4.6.3) – The OPG CBA concludes that none of the conditions associated with non-use benefits apply in this case. This finding is contrary to correspondence from the CNSC to OPG dated October 7, 2008 where the sustainability of local northern pike and brown bullhead populations was raised as a concern. The economic magnitude of this oversight will depend on the level of ecological risk and the social importance assigned to the existence these local fish populations.
  • Sensitivity Analysis (section 4.8) – The OPG CBA includes the results of a limited sensitivity analysis (i.e., for only two parameters, namely, discount rate and end of the PNGS operational life. The sensitivity analysis that is done is fine but this analysis does not include a number of parameters for which the values are uncertain and thus provides limited insight into the upside and downside risks associated with the alternatives.
  • Risk and Uncertainty Analysis (section 4.9) – A comprehensive, systematic uncertainty analysis is rejected on the basis of the results of the sensitivity analysis and the stable ranking of the alternatives. The decision to not use Monte Carlo analysis for this CBA is problematic. Whether the results of a systematic Monte Carlo analysis would alter the ranking of alternatives cannot be stated; nonetheless, the results would provide valuable insights for decision-makers when they are balancing the upside and downside risks of the alternatives.
  • Distribution Analysis (section 4.10) – The OPG CBA does not include a distribution analysis (i.e., an analysis of the distribution of costs and benefits among different stakeholders). Distribution analysis technically, is not part of a CBA. However, a well conducted CBA provides much of the critical information for producing a distribution analysis. Furthermore, the distribution of costs and benefits is a central consideration in many public policy decisions including this case.

These deficiencies need to be evaluated individually and in aggregate in terms of their potential impact on the selection of the preferred CWIS mitigation alternative. Some deficiencies can be resolved fairly easily (e.g., missing authorities) whereas others are difficult to resolve after the fact (e.g., Monte Carlo analysis). We recommend applying the following rule of thumb for deciding to what extent these deficiencies should be resolved; namely, demand further analysis up to the point that the costs of further analysis (i.e., costs in terms of time, financial resources, etc.) is greater than the expected benefits of better information (i.e., benefits in terms of a better decision, greater public acceptance, etc.). OPG is in the best position to evaluate the costs associated with addressing these deficiencies and the potential benefits. For this reason, OPG should be asked to respond to these deficiencies and to justify how they propose to respond to each. CNSC should evaluate OPG's response to decide what further analysis to require.

The OPG CBA and this review will provide valuable direction for future CBAs. For this reason, extra effort should be invested in resolving as much as possible the deficiencies identified in this report.

Top of page

RSP-0262 – Feasibility of reference condition approach for uranium mines

The reference condition approach (RCA) was developed as an alternative design that is intended to overcome the difficulty in interpreting natural influences (e.g., physical habitat features) and effluent influences with respect to community composition. An additional advantage of the RCA is that is characterizes the background variability, unlike site-specific BACI-style designs, and allowing the opportunity to put into context the results from highly powered site-specific control-impact or BACI designs. The RCA typically involves the collection of reference data from several locations (often dozens to hundreds) that vary in underlying physical conditions such as upstream catchment area, stream slope, surrounding geology, water depth, etc.

The monitoring programs that have been developed for the uranium mining sector in the Athabasca Basin of Northern Saskatchewan present a unique opportunity. With multiple mines each collecting data from reference locations, the combined reference data may be suitable for application in an RCA context. The objective of this report is to present the results of a preliminary analysis of the existing reference data to determine the feasibility of implementing the RCA as a monitoring construct for the Athabasca Basin.

Reference condition models can be developed using either regression- or classification-based techniques. Classic RCA programs such as those in the UK (RIVPACS), Australia (AUSRIVA) and Canada (BEAST) use classification-based methods, which have the disadvantage that they lead to potentially erroneous conclusions of impairment for sites that have a naturally different physical configuration than is contained in the reference database. Regression-based methods overcome that error, but are less used.

A reference condition model was developed with lake benthic data from northern Saskatchewan. Lake size (area in ha), lake depth, sample depth, number of grabs comprising a sample and season were used a predictors of total benthic abundance, richness, Simpson's Diversity and Bray-Curtis distances to the median reference community. Unexplained variation in the models was used as the measure of natural variability. The reference database was partially limited because it did not contain lakes larger than 1,000 ha that would be required in order to provide a reference for some of the larger exposure lakes.

The reference models were used to assess each sample collected from a lake exposed to effluents from uranium mines, with deviation from the predicted index values standardized to z-scores. 

The use of the regional reference data, therefore, allows practitioners to put into a broader perspective, the variations that might be observed from comparison of an exposed lake to a site-specific control.

There are few limitations to using reference-lake models for future assessments of benthic community condition for uranium mines in northern Saskatchewan. Further data compilation is required to result in a complete set of template data for all reference lakes. Models can be easily operationalized into Excel work books that can be distributed to industry or government partners for individual use. Models may not need to be updated annually, unless there is an indication of monotonic change in reference conditions over time.

Lake area, water depth, number of grabs and season explained significant variation in either core metrics or multivariate indices of benthic community composition in reference lakes. Standardized residuals from general linear models were generated for each sample for each of six indices of composition including two multivariate metrics. Taxa richness (number of taxa) was the most consistent metric in producing large negative z-scores for exposure sites, implying that many of the exposure lake samples had fewer taxa than predicted for a reference condition. Simpson's diversity index values were never higher than the expected reference condition, and were somewhat frequently less than the expected range of reference values, particularly in D-Zone Pond, East Boomerang Lake, Sink Reservoir and unknown lake. Bray-Curtis distances to the average reference community for samples from exposure lakes were generally not shorter than what was observed in reference lake samples and were sometimes greater. Samples from D-Zone Pond, Hidden Bay Deep and Seru Bay were always in excess of the normal range of expected reference Bray-Curtis distances, while samples from unknown lake, Upper Link Lake and Vulture Lake often produced Bray-Curtis distances in excess of the normal range. In contrast to the conventional core indices of community composition, the z-scores based on multivariate CA Axes 1 and 2 were more variable in their response in exposure lakes relative to the reference lakes.

An example assessment was provided for Aline Lake, part of the Cigar Lake Project with Cameco. Aline Lake tended to have higher total abundances than both Mad Dog Lake and Lake B, higher richness than Mad Dog Lake, lower Simpson's diversity than Mad Dog Lake, higher Bray-Curtis index values than Lake B, and lower Bray-Curtis index values than Mad Dog Lake. Those observations of potential differences would be used as evidence of effluent-related effects in a conventional EEM program. Variations in the standardized indices of composition for Aline Lake fell within the expected range of z-scores for reference locations, and in most cases were within ± 2 SD (i.e., the estimated 95 percent region, or normal range). This graphic analysis of the Aline Lake z-scores illustrates the potential benefit of having multiple reference locations against which to judge observed “effects”.

Z-scores computed for taxa richness, Bray-Curtis distances and multivariate ordination axis scores varied in relation to concentrations of metals and radionuclides in the sediments of reference and exposure lakes. Relationships with metals/radionuclides in sediments were generally similar in reference and exposure lakes implying a natural background influence of metals and radionuclides on benthic community composition.

This analysis demonstrated the relative simplicity of developing reference condition models for lake benthic communities in northern Saskatchewan. Reference models can be developed for any index of benthic community composition, and using any physical or chemical variable considered a template variable that has an underlying influence on the biology of benthic organisms. The analysis carried out here for each individual index of community composition involved development of a general linear model to predict reference sample variations on the basis of season, lake size, water sampling depth and number of replicate grabs. The residual variation from the overall model was then used to standardize deviations from an expected or hindcast reference condition. This assessment, to keep the analysis relatively simple, made a major assumption: that residual variation (background noise) in indices of composition did not differ among lakes of different sizes, between different sampling depths. Had separate analyses and ordinations been conducted for deep-water lake habitats, for example, a different assortment of benthic taxa would have been used in the multivariate portion. It is possible that with a different assortment of taxa that more subtle effects would have been observed in deep-water samples, for example, but that would require further investigation.

Any RCA implemented in northern Saskatchewan will have to consider how many reference lakes to include in the overall design, sampling season, sampling depth and lake size, as each of those factors will influence either the kinds of benthic organisms found in sample, or the ability of the program to statistically detect an effect. A minimum of 20 reference lakes should be identified as potential monitoring locations, while sampling in those 20 lakes should be carried out at different water depths, in different habitats and at different times of the year, in order to capture the conditions that would be found in exposure lake samples. It is probably not necessary to sample reference lakes every year, but annual sampling might consider a kind of panel design wherein a fraction of the reference lakes are sampled every year.

Multi-stakeholder approaches benefit government agencies because the programs are partially or wholly funded by industry. Agencies further benefit from having implemented a program that is considered state of the art, and that has the potential to explore regional impacts, not just site-specific impacts. Industry benefits from the RCA programs in three ways. First, because they are working with the agencies, there is certainty in the acceptance of methods, results and conclusions. Second, there is the potential to share costs and ideas. Third, the RCA data reduce the likelihood of erroneous conclusions of impact because the regional reference data provide a more realistic characterization of the magnitude of background variability.

Top of page

RSP-0263 – Mechanisms of radium-226 mobilization from flooded uranium tailings

A common practice for the uranium industry is to use secondary sulphates to sequester 226Ra for long term immobilization. Unpredicted and somewhat rapid increases in 226Ra concentrations have been measured at certain uranium mine waste disposal facilities after long periods of predictability and/or stability. The migration of 226Ra in the environment is of significance due to its long half-life (1,600 years) and a high degree of radiotoxicity and bioavailability. The release of 226Ra has been the subject of a few studies. It has been hypothesized that 226Ra concentrations are affected by decreasing redox conditions by the direct microbial reduction and dissolution of Ra-bearing minerals, such as radio-barite or by indirect effects, such as the microbial reduction and dissolution of Ra-loaded sorbent minerals (e.g., iron and manganese oxyhydroxides). We investigated whether 226Ra remobilisation was anecdotal or observed across mine tailings in Canada. A comprehensive database of Ra-226 tailings porewater and solids data from different uranium mine tailings facilities was developed along with ancillary concentrations such as pH, Eh, sulphates, sulfides, calcium. iron, manganese, and barium. Results from this study indicate that there are indeed zones of 226Ra remobilization scattered among some of the tailings, in which radium remobilization to pore waters is clearly taking place. However, when the data are pooled, it is apparent that radium concentrations in the waters at most of these sites are actually decreasing (or at least holding steady) over time. Comprehensive information on physical, chemical and biological interactions between radium and the surrounding environment needs to be evaluated and advanced. This is required to develop predictive numerical models of the key processes controlling the fate of radium in U mine tailings. The results of this ongoing research program should allow to interpret and form a conclusion about how site conditions affect the long-term behaviour of 226Ra in uranium mine tailings and identify appropriate management strategies to prevent the remobilization of 226Ra.

Top of page

RSP-0264 – Non-linear modelling and analysis of structural impact

This report describes the development of three LS-DYNA models for the blind prediction of the three experiments comprising the IRIS_2010 Workshop. The experiments consisted of three missile to target impact scenarios, one with target perforation and the other two without perforation. The missiles consisted of two deformable cylindrical projectiles for the non-perforated targets and one essentially non-deformable cylinder for target perforation. The targets were all reinforced concrete slabs of differing properties, both concrete and reinforcement, and dimensions, e.g., thickness.

The simulations results significantly under predicted the concrete target maximum and residual displacements for the two non-perforation impact tests, i.e., by about 70 percent for the MEPPEN test and 85 percent for the VTT Flexural test. It was demonstrated that just changing the Winfrith concrete model (MAT085) to the MAT072R3 concrete model doubled the predicated maximum slab displacement. It is possible that other model changes, i.e., calibration, could improve the displacement comparisons. Comparisons of the missile results for these two impact cases is dubious as the unspecified strain rate effects for the missile material remain an open question.

In stark contrast to the rather poor comparison of the impact results, the simulation results for the VTT punching test are quite good, at least for the missile residual speed. Using the proposed relative error global indictor, the simulation results were the best of the eight participants that predicted target perforation. The predicted damage to the VTT punching targets was a mixed result with a significant under prediction on the front face (spall) but good correlation on the blackface (scabbing). Of course, such damage estimates are highly subjective.

The report concludes with some recommendations regarding additional workshop simulations comparisons and improvements in the conduct of the IRIS_2010 Workshop.

Top of page

RSP-0265 – Mortality (cancer incidence) among Canadian nuclear power industry workers exposed to tritium

There is growing interest in potential health effects from tritium related to the expansion of fusion technologies, the use of tritium in the pharmaceutical industry and possible ground water contamination around nuclear facilities. In Canada, heavy water moderated reactors such as the CANDU are routinely used for generation of nuclear power and are a potential source of occupational exposure to tritium for workers operating these reactors. Most studies of occupationally exposed workers do not report doses from tritium and no study to date has reported risks specific to tritium exposures. We used the data from the recent re-analysis of the Canadian cohort of nuclear power industry workers and new information on socioeconomic status to analyze mortality in this cohort from 1956 to the end of 1994 with special attention to risks from exposures to tritium.

Tritium doses were generally low and had no effect on mortality from solid cancer or any individual type of cancer. Addition of new information about SES did not materially change risk estimates from the recent re-analysis (Zablotska, 2010). In this report we estimated an excess relative risk per Sievert of solid cancer at ERR/Sv=1.77, 95 percent CI: -0.42, 5.30. When tritium was not included in the analysis, risk estimates for solid cancer were 25 percent higher (ERR/Sv=2.21, 95%CI: -0.29, 6.24). Because confidence intervals overlapped, we conclude that the results for solid cancer mortality were not materially changed whether or not tritium was included in the total radiation dose.

No increase in risk of any cancer was observed among ~30,000 nuclear power industry workers from Québec Hydro, New Brunswick Power and Ontario Hydro and the entire increase in risk of solid cancer observed in the whole cohort was due to increased risk among AECL workers. The observed increase in risk of solid cancer deaths among AECL workers could not be attributed to tritium, as AECL workers had generally very low tritium exposures (0.64 compared to 4.70 millisievert for AECL workers and all other nuclear workers, respectively). 

There are no published studies with estimates of tritium-specific risks. Tritium doses to internal organs came from internal exposure to tritiated water so it seems reasonable to include tritium doses in the whole-body doses in all future analyses of this cohort. Further investigation of the group of early AECL workers with unusually high risks of solid cancer due to whole-body exposures is necessary to ensure completeness of their radiation dose records from early exposure years and to improve our confidence in the results of the study.

Top of page

RSP-0267 – Assessment of seismic input and soil structure interaction for deeply embedded, large Foundations

This report describes in some detail work performed over a short period of time (eight months) that deals with some issues of the soil structure interaction modeling and simulations for nuclear power plants. In particular, during this short period of time, effort was dedicated to investigating the following issues:

  • review of techniques and simulation tools used for soil structure interaction (4.1)
  • comparison and critical evaluation of time domain and frequency domain techniques and simulation tools
  • evaluate the amplification of seismic input defined on bedrock due to one dimensional (1D), two dimensional (2D) and three dimensional (3D) effects and assess the importance of 2D and 3D amplification effects comparing to 1D amplification effects
  • assessment of the importance of 1D, 2D and 3D seismic input and amplification effects for massive NPPS structures, with special attention to rocking and torsional modes
  • illustrative numerical examples for analytical procedures, soil and structural modeling, soil structure interaction effects in 1D, 2D, and 3D

In addition to the above issues, effort was dedicated to providing the CNSC with recommendation for analytical procedures and structural modeling criteria, and documenting findings, recommendations and conclusions.

Top of page

RSP-0268 – Loading of steam generator tubes during main steam line break

Tubes in shell-and-tube heat exchangers vibrate when subjected to a cross-flow of shell-side fluids. In order to minimise these vibrations, the tubes are supported at various points along their length and, typically, small clearances between the tubes and their supports exist to allow for thermal expansion and manufacturing assembly. The motion of the tubes in these loose supports may result in fretting wear due to impacting and sliding contact. This wear can reduce the tube wall thickness and therefore the margin of safety against tube failure. Such problems are especially problematic in nuclear steam generators because the tubes represent the boundary between the irradiated primary side coolant (deuterium in CANDU reactors) and the secondary side coolant. The loss of the integrity of this boundary can lead to an escape of primary side coolant from reactor containment and is therefore unacceptable.

In the event of a main steam-line-break (MSLB), the heated pressurised secondary side coolant (water) in the steam generator undergoes rapid depressurisation, resulting in a high velocity two-phase steam-water flow out of the steam generator. This rapid blow-down phenomenon may induce potentially dangerous transient loading on the tubes in the U-bend region of CANDU steam generators due to a substantial rise in the pressure drop associated with sudden liquid flashing. The risk is exacerbated when the tube wall thickness has been reduced due to fretting wear. This transient and the associated tube loading are very difficult to predict.

This report presents the results of an experimental study to develop a better understanding of the transient loading on steam generator tubes in the U-bend region of CANDU steam generators caused by a MSLB. A purpose built experimental rig was designed having a fluid reservoir with a cross-sectional area of about 150 mm2 and a large vacuum reservoir to provide for an expansion ratio of about 60:1. The working fluid was refrigerant R-134a, which boils at about standard temperature and pressure conditions and provides for reasonable scaling to full-scale CANDU conditions. The vacuum tank and liquid reservoir were separated by a rupture disc, which opens in a few milliseconds. Measurements were taken for dynamic pressure and temperature upstream and downstream of the rupture disc and experiments were carried out for various conditions of initial fluid levels with and without the tube bundle test section. High-speed video was used to study the details of the blow-down. An effort was made to measure the tube bundle loading but the shock loading to the entire experimental rig due to the blow-down overwhelmed the load cells designed to capture the tube loading.

The report presents the details of the experimental rig design and instrumentation, and all the results of the commissioning tests and experimental runs.

Top of page

RSP-0269 – Response of a large foundation mat resting on a layered medium to nonvertically incident waves

The objective of this initial study is to consider the response of the large reinforced-concrete lower mat (150x100x1.5m) for a base-isolated nuclear power plant complex subjected to different types of non-vertically incident waves. The emphasis is on an understanding of the physics of the problem and on the development of general tools of analysis. In this initial study, the upper mat and all superstructures are not considered.

Top of page

RSP-0270 – Comments on existing AECL documents used in the seismic evaluation of the NRU facility and recommended acceptance criteria for a current evaluation of the seismic adequacy of the NRU facility

This report contains reviews of previous seismic analysis of the NRU facility performed by AECL in the 1995–2002 timeframe. These previous review analyses used a 1000 year earthquake return period with a standard evaluations elastic response spectral shape as defined in CSA N 289, modified to reflect mean, rather than mean plus one standard deviation ground spectral acceleration factors by dividing the elastic spectral acceleration by 1.2. This elastic spectral was reduced by division of spectral values by an inelastic energy absorption factor, Fµ as recommended in USNRC NUREG/CR-0098 by Drs. N. Newmark and W. Hall in 1978. The Fµ value used was 3.0.

In this report a newer approach to the seismic evaluation of the NRU facility is used which assumes the NRU facility operation is for 10 years from 2011 to 2021. This new evaluation is based on the USNRC Regulating Guide 1.208 and ASCE 43-05 Standard which use a Uniform Hazard Response Spectra for a 500 year earthquake return period which is consistent with the 10-year future operating life of the NRU facility.

This report was peer revised by Dr. Robert Kennedy who took exception to the use of  a 500-year return period evaluation basis earthquake as well, or the use of an Fµ factor as large as 3.0. He also objected to the Fµ factor being applied to the response spectra rather than elastically computed seismic forces, moment or stresses in individual member or sections of the building structures. He recommended that the procedure contained in ASCE 43-05 using a starting point of a 2475 year return period site UHRS modified only to reflect that this was an evaluation of an existing facility rather than a new facility. His result was an evaluation basis earthquake return period of approximately1200 years.

Independent of which approach is taken, Dr. Stevenson has recommended that a nonlinear static push over analysis be performed on the reactor building and overhanging structures to determine instability level of the seismic load results. He also recommends high frequency (fundamental frequency greater than 15.0 Hz) components be reevaluated for the higher acceleration associated with the UHRS input, as shown in Figure 3.1 if the 1200 year return period option is selected.

Date modified: