Very good, sir.
First of all, I think I have to outline what is meant by “didn't work”. To get a licence to build and operate a reactor, one must submit calculations to the regulator, the CNSC, that predict precisely how that reactor will behave. These calculations will show, amongst many other things, what is expected to happen when you want to increase the power of the reactor.
Now, forgive me if I stray into a bit of reactor physics, but I'll do my best to make it straightforward.
To start a reactor, control absorbers, made of a material that absorbs neutrons, are slowly pulled from the reactor in a series of small increments, at incremental distances. The density of neutrons in the core is measured all the time. Initially it's a very small number just due to the spontaneous fission of U-235, which is naturally in the reactor.
Each time the absorbers are pulled out a little further, the number of neutrons increases and then dies back again. Eventually, as you keep pulling the reactor control rods out, the number of neutrons in the core, instead of dying back, equalizes out and is at a steady number. At that point, when the number of neutrons is at a stable level, you have basically a self-sustaining critical reaction. The number of neutrons that are being born in the fission process is exactly being equalled by the number of neutrons that are being either captured by the absorbers, captured by other material in the reactor, or captured by another uranium atom and fissioning. The number of fissions in each generation remains the same. We have a stable population of neutrons.
The core, at this point, is still at a very low power. To increase the power, the control absorbers are pulled out again by small increments, and the number of fissions is increased. You've taken some neutron absorber material out of the core so the fissions increase, and the power of course increases. When the desired power is reached, the absorbers are pushed back in the core to bring the core back to a stable state again in which the number of fissions in each generation is exactly the same.
Now we come to the nub of AECL's problem. A desirable feature of any reactor is that as the power of the reactor goes up, the reactivity--just how reactive the core is--goes down. Let's illustrate this in practical terms. If you want to increase power, let's say, from two megawatts to five megawatts, you pull the absorbers out, let's say, ten millimetres. Those aren't the exact numbers, but we'll use those numbers for the sake of illustration. The power increases, and then as you approach the new power you push the absorbers back in again to stabilize the reaction, as we mentioned. You expect to see the absorbers go back in a lesser distance, say about five millimetres for illustrative purposes. What that says is that the volume of the core is slightly less reactive. You need a bit more volume to produce more power. But you are pushing the control rods back in at a slightly higher point than when you started. You pulled them out at 10 millimetres; you've pushed them back in five millimetres, and the reactor stabilizes or is expected to stabilize again at this new absorber height.
That is what is called a negative power coefficient of reactivity. That means you have a slightly less reactive core at the higher power. What that does is make the reactor easier to control.
AECL designed the MAPLE reactor to have a small negative power coefficient of reactivity, and all its design calculations showed that it did.
When a reactor is first commissioned, an operator has to demonstrate to the CNSC that the reactor will behave in reality exactly as it was predicted to behave in the analysis. When the commissioning tests of MAPLE 1 were first done in 2003, there was a surprise. Instead of the control absorbers stabilizing the reactor at a position slightly farther out of the core, as we just had a look at, once the power had gone up, they in fact stabilized the reactor slightly farther in. In other words, a slightly smaller volume of the core was producing slightly more power. That is termed a positive power coefficient to reactivity.
The difference was quite slight; it's just a few millimetres difference in rod height. The important point, though, was that as far as the CNSC and indeed AECL were concerned, the reactor was not behaving as it was predicted to behave. It was also slightly positive instead of slightly negative. But the important point here is that the behaviour didn't match the predictions.
In a nutshell, the reactor behaviour observed was slightly different from that predicted. And in the eyes of the CNSC and indeed in AECL's own eyes, not being able to predict the behaviour with a high degree of precision really is not acceptable. And as we noted, it is desirable to have a negative power coefficient, as any increase in power is slowed down by the negative feedback. A positive one can be acceptable, provided it is small.
Now, the change in core reactivity with power arises from many different factors, for example, the temperature and density of the moderator and the coolant, and from fundamental properties of the fuel itself. And that is a key result of the physics calculations that AECL was doing.
When the surprise arose, of course, the CNSC stopped the commissioning of the reactor until some explanation were forthcoming.
Now, what did AECL do about it? They did carry out a very detailed analysis of all the possible causes of this observed behaviour, using a panel of their most experienced staff and outside help, and they identified about 200 potential factors, of which about four or five were particularly likely.
They also asked the Idaho National Laboratory in the U.S. to do an independent prediction of the behaviour of the MAPLE core. The Idaho National Laboratory employs some of the very best reactor physicists in the world, and they also have access to the most up-to-date calculational methods. They came up with precisely the same predictions as AECL did; AECL's physics calculations, in essence, were consistent with the world's best physics calculations.
AECL carried out a whole series of tests, and has been doing so in the last two to three years, on the reactor itself, with the CNSC monitoring every step, to show what the contribution was from each of the factors they and other researchers thought were likely to be the source of the problem. The tests showed that some of those factors were indeed contributing to the positive power coefficient, but not enough to explain the whole effect they were seeing. The last set of tests in April of this year showed that the last factor being tested was not contributing at all to the anomaly that had been seen.
AECL's management, as far as I can see, were left with a technical problem for which a solution was not immediately apparent. They had put several hundred skilled engineers and scientists on the task, as well as many external reviewers, without finding the specific cause of the problem. Again, it was not so much that the coefficient was positive rather than negative; it was that they could not satisfy the regulator, the CNSC, that they understood the root cause of the problem.
To solve the problem would likely need the development of new fuel that would be designed to have a definite negative power coefficient of reactivity. That's a task that takes several years and several million dollars.
How could AECL be faced with such an unknown today? The MAPLE reactor is unlike any other reactor. It's very small—about the size of a garbage can. It's quite small, and has a combination of highly enriched fuel in the “targets” that are to be harvested for their isotopes, and a combination of low-enriched and depleted uranium fuel to drive the core. It is a very non-homogenous core. This small size and the unusual reactor physics seem to have introduced a very sensitive interaction between the dimensions of the core, the fluid mechanics of the core, and the reactor physics that has not been observed before.
Thank you, Mr. Chairman. I'd be pleased to answer any questions the committee has.
