In a recent Spectator article Mr Rishi Sunak complained that the scientific advice that was used in COBR and SAGE was of poor quality in that it did not clearly outline what was known and what was assumed and the impact of assumptions on the output.
‘I was like: “Summarise for me the key assumptions, on one page, with a bunch of sensitivities and rationale for each one”,’ Sunak says. ‘In the first year I could never get this.’ The Treasury, he says, would never recommend policy based on unexplained modelling: he regarded this as a matter of basic competence. But for a year, UK government policy – and the fate of millions –was being decided by half-explained graphs cooked up by outside academics.
Would the nuclear industry fall into this trap if advising people to shelter or can we do better? Can we explain why we might ask people to stay in their home during a nuclear accident that is releasing radioactive material to the atmosphere? Here is my attempt.
I think that one page is optimistic if briefing those new to the subject. Covid was a little different in that a couple of weeks into the crisis COBR and their customers would be better up to speed. So here is my attempt at a one page brief for a hypothetical situation and introductory annex.
Advice brief
(This is what the one-page advice summary may look like)
Advice given to STAC/COBR on 12/12/22 with regard to public protective actions
The default automatic public protective action of recommending everyone in the DEPZ take shelter was promulgated using the automatic phone system (reach estimated to be 50% of those in the DEPZ) as well as on broadcast and social media starting at 10.15 on 12/12/22 following the raising of the alarm by the Operator at 10.00 on 12/12/22.
Evidence suggests that compliance has been [good/indifferent/bad].
The situation at 11.30 is that the release is continuing but gradually reducing as temperatures and pressures reduce. Attempts to stop the leak are also continuing. Plan A has an estimated 50% chance of succeeding by 12.00 and a 70% change of succeeding by 12.30. Independently Plan B has an estimated 90% chance of succeeding by 13.00. Irrespective of these plans the release is expected to end by 14.30.
The dose estimates are based on the NAME computer model with best estimate parameters, a source term estimate from the operator, weather forecasts from the Met. Office and the advice based on the Lower Emergency Reference Level of dose for evacuation from the PHE (now UKHSA). Deviations of the actual dispersion from the modelled dispersion may lead to different dose distributions. In particular, deviation of the weather conditions from those forecast would change the distribution of potential dose off-site.
It is assumed that shelter reduces dose by 40% compared to being outside. This is the default recommended by PHE (now UKHSA). Delays in achieving shelter or shelter in a less capable building will reduce the gains in terms of averted dose. Conversely better than expected performance will increase the dose averted.
Considering that the most likely outcome to be the success of plan B at 13.00, it is estimated that sheltering if applied immediately, or is already in place, will avert dose equivalent to the lower ERL for shelter to 2.3 km downwind. This reduces to 1.8 km if plan A succeeds by 12.30.
Considering the weather forecast, particularly the wind direction, we recommend that Sector B be asked to shelter out to 2.5 km, or somewhere beyond, because estimates show that it is in this region that the lower ERL for shelter is likely to be exceeded.
Thought should be given to sheltering sectors A and C to the same distance in case the predicted direction of plume travel proves to be slightly inaccurate.
The other sectors could be asked to remain in shelter as a precaution against the dispersion estimates (notably wind direction) being very wrong or released from shelter if there is confidence in the dispersion estimates and the benefits of doing so outweigh the risks.
If the accident develops in unexpected ways leading to an increase in the release rate or a longer duration release this provision may be found inadequate after the event. Conversely, if the release is terminated sooner, they may retrospectively be thought of as excessive.
Subject to the caveat about the release not being significantly greater than currently estimated, there is confidence that the off-site doses will not exceed the Lower ERL for evacuation therefore evacuation is not recommended as a prompt protective action for the public.
Annex
- Radiation is to be respected not feared.
Radiation has many positive benefits to modern society particularly in medicine, engineering and, of course, power generation. I am aware that post-accident we probably wouldn’t include this automatic PR padding.
We know that radiation in large enough doses (over 1,000 mSv) over a short period of time leads to injury (known as deterministic effects). The literature contains enough gruesome pictures of radiation burns and stories of injury and death following radiation exposure to take this as fact.
There are also longer term effects, such as the potential for cancer induction, to take into account (Stochastic effects). The probability but not the severity of these effects, which may take years to express, are believed to be proportional to the radiation dose. ICRP-103 states that “in the case of cancer, epidemiological and experimental studies provide evidence of radiation risk albeit with uncertainties at doses about 100 mSv of less”. It is assumed, but generally also widely accepted, that in the range from 1 mSv to 100 mSv the risk of harm in this manner is directly proportional to dose.
A lot of research has been undertaken and the currently accepted model (from ICRP Publication 103, 2007) assigns detriment-adjusted risk coefficients (see note) of 5.5 10-5 per mSv for cancer and 2.0 10-6 per mSv for heritable effects in the whole population (including infants, children and adults). This means that each mSv of dose increases your risk of cancer by 5.5 x 10-5 which is low compared to your overall risk of cancer (Cancer research UK is claiming that “1 in 2 UK people will be diagnosed with cancer in their lifetime”) but arguably worth avoiding if possible and cost effective.
Note:
Detriment-adjusted risk is defined as “The probability of the occurrence of a stochastic effect, modified to allow for the different components of the detriment in order to express the severity of the consequence(s)”. It is an attempt to summarise in one number the harm that radiation may do to you in the years after exposure.
Radiation Protection is a well-developed science and our ability to reduce radiation doses is good and improving. In an accident (and in planning for accident response) we aim to reduce dose well below the thresholds for deterministic effects (aiming to keep doses below 100 mSv) and the reduce dose further, if possible, to reduce the risk of stochastic effects.
Given that there is an average background radiation dose of about 2.7 mSv to the UK population, putting significant effort into reducing additional doses that are much below this range is rather pointless.
- You may receive a radiation dose in a nuclear accident.
Highly radioactive material is an unwelcome by product of the energy generated in nuclear reactors. Great effort is taken to ensure that this dangerous material is kept within several layers of containment and is unlikely to get into the environment in damaging quantities.
However, in the event of a nuclear accident that releases radioactive material (dusts and gases) to the air the radioactive material will drift downwind and spread out as it goes. If we know the rate at which radioactive material is being released and the weather conditions, particularly the wind direction and speed, we can estimate the concentrations of radionuclides in the air at any point downwind as a function of time.
From this we can estimate the radiation dose that a person standing at that point will be exposed to. This radiation dose is composed of dose due to inhaling the radioactive material (inhalation dose), dose due to being near the radioactive material (cloud dose and ground dose) and dose due to eating the radioactive material as contamination in food and water (ingestion dose).
We can also estimate the dose averted (saved) if we move that person out of harm’s way or put them in a shelter at some time before or during the release.
Inhalation dose is usually the dominant component. Inhalation and cloud dose occur during plume transit. Ground dose (due to radioactive material deposited on surfaces) and ingestion dose (due to contaminated food and drink) occur for some time after plume transit depending on circumstances. It is possible for deposited activity to be kicked up in the air again, but the doses (resuspension dose) resulting from this are a very small fraction of the original dose (see box).
Box
When a plume of radioactive dust travels across an area, a fraction of the activity is deposited onto the ground. Of this fraction, a further fraction is resuspended. Taking plausible values for these fractions from the literature (1 x 10-2 Bq.m-2 per Bq s m-3 for deposition and 0.75 Bq s m-3 per Bq m-2 for resuspension over a year), it follows that the dose over one year from resuspension is likely to be about 1% or less of the plume transit dose (1 x 10-2 x 0.75 = 7.5 x 10-3 Bq.s.m–3).
- Sheltering, evacuation and stable iodine may be used to reduce your radiation dose in certain circumstances.
If you are given adequate warning of a nuclear accident, or the release might be expected to continue for several hours, there are simple steps that can be taken to reduce the potential radiation dose to the public.
Going into a building (shelter) reduces the inhalation dose because, for a period of time at least, the airborne concentration inside the building will be lower than that outside. The building also provides physical shielding that reduces cloud and ground shine.
Leaving the area before the plume arrives or before the release finishes (evacuation), reduces dose by removing the person away from the hazard. Evacuation is more difficult to organise and achieve than shelter and can be very disruptive. Some vulnerable groups can be harmed in evacuations (Fukushima experiences) and psychological impacts can be significant (Chernobyl and Fukushima experience). Because it is more difficult than shelter, we initiate evacuation at a higher dose saving threshold than shelter.
Stable iodine tablets can be used to saturate the thyroid gland to prevent it taking up radioactive forms of iodine that might be released in a reactor accident. This reduces thyroid dose and therefore the probability of thyroid cancer in later years.
The latest government sponsored advice on these “Protective actions” is given in CRCE-049. This justifies and restates Emergency Reference Levels which it defines as the “dose criteria that apply to the justification and optimisation of sheltering-in-place, evacuation and administration of stable iodine”. ERLs come in pairs. The Lower ERL gives the value of averted dose below which a protective action is unlikely to be justified. The Upper ERL gives the value of averted dose above which the protective action is strongly recommended. Between the two values is an area where the difficulty of implementing the protective action on the day comes into play in the decision-making process.
- We advise people in some areas to shelter because we think the dose saving in those areas makes it worthwhile and do not advice people to shelter in those areas where the costs outweigh the benefits.
In the event of a nuclear accident urgent attempts will be made to reduce and stop any release of radioactivity to the environment. Judgements will be made about how long the release will continue and how the release rate will vary with time.
This information and forecasts of the local weather conditions will be used to estimate the dispersion of radioactivity in the environment and its dose implications to members of the public in different locations off-site. These dose estimations, and estimations of the dose that could be saved by implementing protective actions will form part of the decision-making process.
Responders with local knowledge and the information collated in the off-site plan will consider the expected dose distributions and the ease at which protective actions could be put in place to determine which areas to include in protective action advice and which areas not to.
Areas subject to protective actions will be those where the avertable dose seems likely to be above the relevant Emergency Reference Level (protective action is likely to do more good than harm) and areas where the advice is given not to implement the protective action will be those where the costs (in the broadest sense) of implementing the protective actions exceeds the likely benefits.
- Uncertainties
They are considerable uncertainties in this chain of argument.
- The detrimental impact of radiation at these levels of dose and dose rate. We use the international consensus models of radiation harm. Source: ICRP.
- The values that should be chosen for the ERLs. We use current national advice which is consistent with current international advice. Source: UK HSA.
- The radiation doses the public may receive in the event. We use models that consider all the data available (situation on site, any measurements that might imply the release rate on site, release prognosis (possibly worst reasonable outcome and best estimate outcome), off-site measurements of radiation and radioactivity levels, weather forecast). These models give answers that depend on the assumptions made in the input data (most sensitive to the source term and weather conditions), the assumptions in the model itself (the physis of dispersion and how it is conceptualised) and the uncertainty in implementing the model on a computer. Source: Operator, UK-HSA and Met. Office and potentially with ONR oversight at each stage.
This data will be sparse, time will be short and considerable professional judgement will be utilised so the dose estimates will be subject to considerable uncertainty.
- The avertable dose depends on how quickly the alert can be promulgated, how quickly and how well the public respond, and how effective the protective action will be. Again, there are some uncertainties in these parameters.
Summary
The impact of these uncertainties is that the advice to implement protective action over a particular area will, itself, be uncertain. A different set of assumptions may lead to a larger area, a smaller area or a different area.
There is a tendency to err on the side of overestimating dose and rounding up predictions so the protective actions are likely to be recommended over a wider range than the underlying science and situational awareness might suggest.
Afterword
The “costs” of short-term shelter are not considered to be great. Initial overstatement of the area that would benefit from shelter would not carry a great cost burden providing it can be withdrawn in a timely manner. Over enthusiastic emphasis on strict shelter (for example, preventing emergency services and essential carers into the area in cases of need) over a wide area and a prolonged period would begin to accrue disproportionate costs.
Evacuation and the subsequent care of a significant number of people has its problems in terms of organisational complexity and resource requirements so more care is required to ensure it is not used indiscriminately.
The taking of stable iodine tablets is not expected to result in many health impacts but if advice is given to take them too soon there may be a need to issue a second round of tablets which has resource and dose-to-responder issues.
The real psychological problems related to nuclear accident seem to result from the stresses of living in a post-accident area with environment contamination leading to social and economic disruption and stress. Management of this is very important and starts soon after the accident but generally after the prompt stage.