Effects of sodium aerosol deposition in LMFRs

Most of fast reactors have to some extent experienced problems related to sodium aerosol depositions. At BN-350 and BN-600 some deposits were observed in the gap between the large rotating plug and the reactor vessel roof, inducing difficulties of rotation. At KNK-II and Phénix, deposits have been found in the control rods between the shielding piston and the guide tube, causing difficulties with their insertion.

At PFR, the possibility of sodium aerosol deposition in those parts of the absorber drive exposed to the primary vessel gas blanket was recognized in the design and an argon purge flow was provided in these areas to prevent aerosol transport. However, it was not completely effective and the removal of small deposits on the magnet faces in the drive mechanisms was a routine procedure at PFR. This led to plant trips on a number of occasions due to absorbers dropping off their magnets during power operation.

The question of difficulties with the PFR control rods has been examined in two phases.

During the first phase (up to late 1985) the results gave no cause for concern. Since 1985, the reactor operated at high power and high temperatures and greater problems were encountered.

It was believed at the beginning that sodium deposition was a main cause and after investigations confirmed this. Particular features to note are the roof insulation region, the weight sensing equipment and the dashpot. Initially, it was thought that the main issue concerning the freedom of rod movement would be bowing caused by neutron induced voidage (NIV) coupled with thermal effects. The computer code Peeble predicted maximum friction of 25 kgf. However actual measured maximum friction was 40 kgf and the distribution of measured friction with rod position was not in good agreement with the Peeble prediction.

PFR had 5 shut-off rods (normally fully raised) and 5 control rods inserted to control power.

The rods were essentially identical B4C assemblies supported by electromagnets (Fig. 12).

On a trip all ten rods dropped. Magnet current, apparent rod weight, rod release time and time of flight were measured by installed instrumentation

At all shutdowns and after plant trips the electromagnet pick-up and drop-off currents were measured. These were the minimum magnet currents at which the absorber could be raised and at which it dropped off after being raised. On the basis of these figures a decision was made on whether the magnet faces had to be cleaned before return to power. If required the drive and magnet assembly were removed by simple bagging techniques and the magnet face was cleaned in an argon purged glove box. The extension rod face was cleaned in situ using commercial “Scotchbrite” cleaning pads, again making use of a simple bagging technique.

After late 1985, when the high power operation started a sudden increase of friction was measured with values going up to 80 kgf. At this time the possibility of sodium aerosols in the cover gas and deposition on the keys and key ways in the upper part of the mechanism were discussed. As an explanation it is suggested that sodium has deposited in the positions indicated in Fig. 13.

FIG. 12. The PFR absorber rods.

Deposits in these positions could give the distributions of friction measured, largely as a result of the detailed design of the keys which are attached to the “Latch/Delatch” tube and which engage in key ways in the extension rod. In 1988 a special glove box was made which allowed examination of the liner tubes and the extension rods. Examination of a number of rods confirmed that sodium deposits were present but in smaller quantities than expected and confined to the keyway of the extension rods. None were found on the liner tube as originally hypothesised. The sodium was soft and easily removed. Although the absence of deposits other than in the keyways was surprising, when they were removed the friction of the restored rods to normal. It took some 40 effective full power days (efpd) of operation for friction levels to begin to rise noticeably.

A difficult problem in operation was posed by sticking in the shutdown systems of KNK-II (Fig. 14).

FIG. 13. Location of sodium aerosol deposits on a PFR absorber rod mechanism. FIG. 14. Schematic representation of the shutdown rods in the KNK-II shutdown systems.

In December 1986, a control rod of the primary shutdown unit for the first time was found to stick while the reactor was shutdown. The cause was found to be sodium aerosols plated out during prior handling steps, when the rod actuating equipment had not been swept with gas.

In December 1988, deposits were found on a rod of the secondary shutdown system; they impaired the mobility of the component, but not the shutdown function. Probably the fact that the primary system had been opened for maintenance purposes a number of times before had caused the quality of the cover gas to deteriorate and thus produced the deposits. On January 1991 the scram at 15% power reactor operation was initiated by a sudden absorber movement after an obstruction in movement had been overcome. This blockage in the primary shutdown system again was caused by depositions in the rod actuating equipment in a phase in which the cover gas quality had been insufficient.

When the quality of the cover gas is insufficient, the sodium in the rod actuating equipment was oxidized to sodium oxide whose dough-like consistency impeded lifting movements of the equipment. This blockage in the primary shutdown system was caused by depositions in the rod actuating equipment in a phase in which the cover gas quality had been insufficient.

Movable parts within the primary reactor envelop, might be exposed to cover gas carrying considerable amounts of sodium aerosols, have to be protected by structural elements like expansion bellows or double seals with cover gas in between.

If narrow gaps between fixed and moving parts within the reactor vessel gas plenum are unavoidable and venting gas has to be applied, measures for the flow and quality surveillance of such venting gas are essential. In case of poor gas quality, caused by impurities like hydrogen, oxygen or methane, a possibility to switch over to clean gas from the liquid gas storage tank has to be installed. Depending on certain circumstances the gap has to be chosen such as to minimize possible convection of sodium aerosols.

Solving the problem of aerosol deposition is a difficult one because it is related to gas convection and to the temperature of the different structures. In the frame of the new project, an important effort is being made for the development of a high performance calculation code.

If needed, design provisions should be incorporated such as tight annular spaces for the rotating plugs, large clearances between moving parts, heating devices to avoid solidification.

Also, gas injections should be limited to a minimum in order to avoid oxide formation and deposits on moving parts. The experience with short-circuit devices for the detection of sodium leaks is good. Nevertheless, an important effort is still necessary to improve their sensitivity and reliability.

One solution is to use metallic (ferritic) reactor roof in which the component penetration are machined, the tolerances are more controlled, so smaller clearances are possible. It gives a further benefit of lower a sodium aerosol and heat transfer from the cover gas.