Jeremy Bernstein in the New York Review of Books:
In fact, to anyone who has been following the Iranian nuclear program, it was almost a forgone conclusion that negotiations with Iran would hit a road block when it came to the so-called IR-40 reactor located in Arak. The “40” here refers to the projected power output of forty megawatts of thermal power. To convert this into electric power involves a cumbersome process. The thermal power, which is generated in the form of energetic fission fragments in the reactor, must be converted into steam to run inefficient steam turbines. Thus much of the original reactor generated energy is dissipated; something like only a third of this power could be converted into electricity. And since one large building alone can use several megawatts of power, it is hard to imagine generating much electricity from a forty-megawatt reactor. Whatever the IR-40’s intended use, it is not to produce electric power. A reactor designed for that purpose—such as the one at Bushehr—produces billions of watts.
Moreover, there is nothing about the reactor’s declared purpose that would require it to be a heavy water reactor. According to the Iranian government, the IR-40 reactor is supposed to make medical isotopes. But a light water reactor would have served the stated purpose just as well—and generate the same amount of power.
What makes the Arak reactor suspicious, then, is the design. To understand this we need to understand how a heavy water reactor works.
More here.