November 21, 1996
In my last presentation, I addressed the issue of the “burden of proof” for the safety of this concept, and I have since added a few more sentences on the record, based largely on the important contribution of Rod Northey last Monday evening.
In short: the governments of Canada and Ontario, after painful experience with an approach that began by investigating possible sites, decided that no potential site would be considered or selected until the concept had been assessed and proven safe and acceptable. This Panel may advise the federal government that that decision was misguided, or that it has outlived its usefulness; but it cannot unilaterally reverse that policy decision.
Instead, you may be tempted, as the SRG was, to adopt unique and ad hoc definitions of “safety” or of “proof”, or to reverse the normal burden of proof. Here is SRG’s “bottom line” conclusion (November 18, last slide):
The concept could, in principle, be implemented safely and effectively.
The AECL post-closure performance assessment methodology is unreliable . . .
This does not mean that the . . . concept is unsafe. [NB burden of proof!] Safety will have to be demonstrated for each specific site.
As helpful as SRG’s input has been to this process, they have not answered the EARP Guidelines Order’s question about the safety of the concept, as you the Panel must.
But whatever beliefs the technical community may share about the state-of-the-art of engineering, geology, and computer modeling, and their adequacy to produce a safe result, parties like SRG and AECB and TAC should not be allowed to use the word “acceptable” in their statements of approval. For “acceptability” clearly means something different from “safety”, and this panel is expressly charged with judging the Concept on both.
What, then, constitutes proof of acceptability, and has it been presented to the Panel? In this context, I would submit that before a concept is proved acceptable it must reflect the values of the Canadian public, and it must be very likely to produce a result that will be widely judged to be acceptable. On both tests, this concept must be rejected as unacceptable, regardless of whether the Panel judges it safe or not. If the case for safety is incomplete and flawed, the case for acceptability, in our view, has actually been disproven on the evidence.
Indeed, you have heard clear evidence — hard, factual, historical evidence, from Port Hope, Meadow Lake Tribal Council, Germany, and elsewhere — that this concept is very likely to produce a result that will be judged unacceptable, divisive, and painful, even if more research resolves the important technical concerns of the SRG and AECB and others. The likelihood of just such a result prompted an earlier Panel under the EARP Guidelines Order to reject a uranium refinery at Warman, Saskatchewan as unacceptable.
Some of the concerns that undermine the acceptability of the concept are partly technical, while others are more philosophical, moral, or political. AECL’s and Ontario Hydro’s research has indicated a strong public desire — shared by Energy Probe and many other public-interest critics at these hearings — to avoid permanent disposal unless meaningful monitoring is incorporated. And AECL has indicated that its technical ability to deliver that monitoring in their planned repository is extremely limited, to be charitable. And with all due respect to AECL’s attempts to prove otherwise, real post-closure retrievability (another apparent requirement for acceptability) seems also to exceed the feasible state of the art. Incidentally, we attribute part of the blame for these unacceptable features (the lack of retrievability and long-term monitoring) to the dominance of AECB’s Regulatory Document R-104 in the development of the concept — a regulation which was developed with only a handful of written comments outside AECB, almost all of them from government and industry, and which clearly does not reflect the values of the Canadian public.
Several of the less technical barriers to this concept’s acceptability are items outside this Panel’s mandate. For example, many intervenors have told you that pursuing nuclear waste disposal while continuing to produce more nuclear waste is unacceptable. Indeed, I recall that Dr. Claes Thegerstrom from SKB in Sweden said so clearly on the first day of Phase One, “The life of SKB would be easier if there was a firm sign that nuclear would be phased out.” He also indicated that the issue of foreign importation of nuclear waste was an important one despite Swedish legislation forbidding it. Canada, of course, has no such legislation or policy. Of course, it is outside your mandate to tell government whether or not Canada should phase out nuclear power, or whether or not Canada should accept nuclear waste from China, or Thailand, or anywhere else. But it is squarely within your mandate to tell government whether or not the concept of deep geological disposal of nuclear wastes is acceptable and, if not, why not.
Another barrier to acceptability is the lack of any successful demonstration of AECL’s stated principles of voluntarism in the history of the Canadian nuclear industry, including, of course, the low-level waste siting task force’s failed attempt to apply those principles. AECL’s pledge of allegiance to voluntarism and openness (in the EIS) is no more a proof of concept acceptability than a pledge of allegiance to the principles of risk reduction would be a proof of concept safety. Acceptability, like safety, must be proven. We doubt that AECL has proven safety, but we must admit that they have tried, and tried very hard. We cannot say the same about acceptability.
In this context, we note two bits of hard evidence that we believe prove the unacceptability of the concept’s approach to voluntarism: (1) Natural Resources Canada, in its new Discussion Paper on Institutional and Financial Arrangements for the Disposal of Radioactive Waste in Canada, has shown in two ways that 1996 is much too soon for Canada to begin an acceptable, voluntary, and open process: (a) The document was apparently circulated for comment to parties within the nuclear industry, but not to public-interest groups with a long-standing interest in the subject; and (b) The document itself enshrines such one-sided “consultation” on an ongoing basis: “[Waste disposal plans, and financial arrangements] should be reviewed and approved by the federal oversight body [i.e., Natural Resources Canada] and the AECB. The review and approval mechanism will need to be carried out in an inegrated manner and be transparent and straightforward to waste producers and owners. [p. 16]”
The second bit of hard evidence concerns AECL’s cavalier treatment of this Panel’s requirements in general, and specifically in the area of explicit exclusion criteria: Many intervenors, and the SRG, have pointed out the desirability of explicit exclusion criteria since the Scoping and Guidelines Phase of this Hearing, and the Panel included several items related to them in their Guidelines document. But AECL has steadfastly and consistently refused to adopt any. Not only could good exclusion criteria help to eliminate the possibility of the concept producing unsafe outcomes (a possibility which has not been developed in this process because of AECL’s blanket refusal to open the discussion); good exclusion criteria are also an essential step, in our view, toward creating acceptability. In this context, AECL’s presentation of the second reference case (in an attempt to prove widespread applicability) has seriously undermined the acceptability of their concept, by openly raising the possibility that the concept could produce a result that virtually all parties consider unacceptable, but which passes R-104, the one regulatory hurdle that AECL has consistently followed, even in preference to this Panel’s requirements.
The second reference case, while proving that AECL could easily satisfy AECB’s R-104 — and thereby satisfy AECL’s own circular, legalistic definition of “safety”! — also further undermines the definition of the concept (something we already found vague enough). Specifically, we are no longer sure what AECL means by “multi-barrier”. We had taken it to mean something like its reactor-safety analogue, “defense in depth”. But in reactor safety, the concept of defense in depth is almost always used to describe the simultaneous provision of redundant safety features, each of which could ensure safety in the absence of the other. But even if “safe” is defined as “meeting regulatory requirements”, the only way the two test cases could pretend to prove safety with this kind of defense in depth is if the concept insisted on the extremely low-permeability geosphere from the first test case and the extremely long-lived container from the second. Clearly that is not the concept this Hearing is being asked to approve, nor would “optimisation” by AECL, or Hydro, likely produce such a result.
[Finally] we want to call the Panel’s attention again to the past, present, and future issue of inertia. I believe we all know that twenty years and $500 million of assuming that the object was irretrievable, unmonitored disposal have created an impressive reality, as have seven years of concept review. In addition to the obvious appeals to “keep the team together”, it is also tempting for all parties to look at that pool of “sunk capital” and to retroactively invent as much meaning for it as possible. The research program simply can’t have been pursuing the wrong goals; the environnmental assessment can’t have been trying to answer an impossible question, or the wrong question, or one that isn’t as meaningful as it seamed to be. These things can’t be true, can they?
That inertia is a major challenge for us today. But if this Panel approves this concept by stamping it “safe and acceptable” — even with volumes of reservations and qualifications — the inertia will grow enormously, and so will the challenge.
[claims of “incredible”containment success, immediately below]
[Background radiation, below]
[ICRP and non-fatal cancers, below]
For many years in the nuclear debate, the discussion about nuclear wastes has boiled down to a simple “dialogue of the deaf”. Nuclear opponents claimed that the enormous toxicity of nuclear waste, and the extremely long life expectancies of some of their components, meant that their containment had to be incredibly perfect for incredibly long periods of time — and that confident claims of success were not credible. Supporters of nuclear energy countered by saying that technology was available or could be developed to achieve the needed containment and confidently demonstrate success. Twenty years and half a billion dollars later, we have learned a great deal about many things, but the fundamental disagreement persists, e.g.:
Vol. 5 (Radiological Assessment), Table 8, p. 80: AECL’s models, including those in the Second Reference Test Case, contain tonnes of inventory of each hazardous radionuclide, allowing only a few grams into the entire biosphere, even after 100,000 years. In other words, we are told to believe that over 99.998% of the remaining radioactive inventory of the high-level waste will still be effectively contained after 100,000 years, even in the second test case, which only marginally meets the AECB risk criterion in the average of the scenario runs. Containment in the original, EIS test case is generally even better. (Believe that, and you believe that the disposal is pretty successful, safe, etc.)
For example, Iodine 129, the dominant contributor to dose rate in the EIS case study, has an extremely long half-life (1.57 x 107 yrs). Furthermore, “about 8% of its total inventory is instantly released from the irradiated UO2 fuel. It has a high solubility and is one of the most mobile radionuclides, generally exhibiting no sorption or weak sorption on the engineered and natural barriers, except for overburden and lake sediment. . . iodine is not significantly delayed in the geosphere. [Vol. 5, p. 83]” Its initial inventory is 3.1 x 104 moles, i.e., 129g x 3.1 x 104 = 400 x 104 g, or 4 tonnes. But despite its rapid transfer through the geosphere, and its near-total lack of radioactive decay, the total amount released to the biosphere in the first 100,000 years is 7 x 10-1 mols, x 129g = 90.3 grams, or about 3 ounces. (That is because almost none of it is predicted to escape from the canisters.) Put another way, only a little over 22 millionths of the initial inventory makes it to the biosphere by the end of 100,000 years! Believe that with confidence, and you also believe that deep geological disposal is pretty successful, safe, etc.; doubt it, and you doubt it.
Vol. 5 (Radiological Assessment), p. 86 and Figure 27 (p. 87) present the Radiological Risk per year, vs. time, for the Second Reference Test Case, “using two different risk conversion factors that differ in their definition of ‘serious health effects’. In 1987, the AECB specified a value of 0.02, where serious health effects are fatal cancers or serious genetic effects occurring to an individual or his or her descendants who would be exposed to the greatest risks (AECB 1987 [R-104]). More recently, the ICRP has recommended a value of 0.073 where serious health effects are fatal and nonfatal cancers or severe hereditary effects (ICRP 1991a [“ICRP-60″]).” The note to Figure 27 explains that the curves “plot the conditional radiological risk, or probability of a serious health effect per year, but use different risk conversion factors. [italics added]”
What AECL does not mention here is that the ICRP risk conversion factor does not actually estimate the “probability of” “fatal and nonfatal cancers or severe hereditary effects”. Rather, the 0.073 is ICRP’s suggested coefficient (after dividing the “best-fit curve” by 2 for DDREF) for fatal cancers and serious genetic effects and only a fraction of serious non-fatal cancers. Including all serious non-fatal cancers (i.e. all but the relatively trivial basal cell carcinoma), as is generally done for the risks of chemical carcinogens, would increase the ICRP’s coefficient to approximately 0.1, or 10% per Sv. (Of course, leaving out the unproven and non-conservative assumption of a DDREF would double either of these coefficients. See Energy Probe’s fourth submission to this Panel, especially the discussion at pp. 7-10, and endnotes 17, 19, and 33.)
The risk curve in Figure 27 on p. 87 peaks at a level a factor of ten below the AECB’s risk limit. Any reasonable combination of the following reasonable corrections would raise the risk curve in Figure 27 on p. 87 over the line of acceptability:
Lower the “acceptable” risk from 1 x 10-6 per year to 1 x 10-6 per lifetime, in line with modern regulations for known or suspected carcinogens. (This change is “worth” a factor of about 76, at today’s life expectancies.)
Reject ICRP’s discounted concept of “detriment” and include all serious non-fatal cancers as “serious health effects”, in line with modern regulations for known or suspected carcinogens. (This change is “worth” a factor of about 1.3, compared to the ICRP curve.)
Reject ICRP’s unproven and non-conservative assumption of a DDREF of 2 for cancer effects, and use at least the “best-fit” coefficient: 0.1 per Sv for fatal cancers alone. (This change is “worth” a factor of 2, compared to the ICRP curve.) Modern regulations for known or suspected carcinogens go one step farther, by choosing a risk coefficient from the “conservative” or “high” end of the 95% confidence interval of the data.
This risk is based on the average scenario. (Cf. Figure 17 on p. 56: the curve that approximates the risk curve on p. 87 is the average.)It is essential that we have a high level of confidence that our maximum risk criteria will not be exceeded, and not just a linear, 50-50 expectation.
Figure 17(a) on p. 56 shows percentile bands for the dose rate to the Critical Group from fission products. Across the top is a horizontal line, explained as follows: “The horizontal line at 3 x 10-3 Sv/a is representative of dose rates from natural background.” This is misleading in a number of ways:
The doses from this facility are all predicted to be in addition to the dose rates the Critical Group will get from natural background radiation.
Showing the dose rate that corresponds to an annual risk of 1 x 10-6 per annum of “serious health effects” — about 1.4 x 10-5 Sv/a at ICRP’s 0.073/Sv, lower at higher coefficients — would clearly indicate that the highest observed value goes above that line at 100 years and stays there until 1,000,000 years. It seems to be AECL’s unspoken policy never to present any horizontal line on any graph which any curves cross!
The figure for natural background radiation is certainly higher than the Canadian average for total natural background radiation, which has been estimated at 2 x 10-3 Sv/a — see, e.g., TAPNS Report #93-V, (16 March 1994), which included AECL’s D.K. Myers as a co-author, and which cites Health Canada as a source.
Since the primary route of exposure to radiation from the facility is via “whole-body” irradiation by internal deposition of radionuclides, it would seem that a more meaningful comparison would exclude the dose from radon gas. The non-Radon dose seems to be more in the order of 1 x 10-3 Sv/a — see TAPNS 1994 for both the estimate and a discussion about the meaningful comparison.
The dose rate from C-14 has dropped from the EIS because of new assumptions about initial inventory and instant release fraction, both based on new experimental data from “Johnson et al 1996” which seems to mean v. 2 of this study. Cf. vol. 5, p. 82.