« Disabling the Dell Inspiron Touchpad on Fedora Core Linux | Main | Sketch of the Analytical Engine Updated »

Friday, February 17, 2006

Lac Fourmilab: Pictures from an Inundation

lac1_2006-02-17.jpg After thinking I had finally gotten everything out of the way so I could do some actual work, I walked into Fourmilab in high spirits on Thursday the 16th and what should I find but 5 cm of water throughout the basement, with the furnace room filled to the brim—about 20 cm deep there. The water had flowed over to the elevator and filled the shaft level to the basement floor, two metres deep, on the order of 18 cubic metres of water there alone. (Click on images to view enlargements.)

The cantonnier from the commune came out with a pump and wet and dry vacuum, and as of around 17:00 most of the standing water and the contents of the elevator shaft had been evacuated. The vacuum is staying the week-end so I can deal with anything that happens.

It was not at all obvious where the water came from, nor is it certain even after a day of investigation and ratiocination. There has been no leak in the water system, and there was no sign whatsoever of an entry point (water stain on a wall, etc.). The water was seemingly clean, but could have been rainwater or melted snow. The evidence pointed to it welling up in the furnace room, but how that much water could have gotten through a concrete floor and walls in less than 24 hours without a big obvious crack was a total mystery. In an attempt to identify the source and avert any further lac2_2006-02-17.jpg disaster, I am inspecting the situation every three hours while the rain and melting snow continues to generate run-off and ground water . The sewage pumps are functioning normally, and there is no back-up of water in the cistern they pump from, nor is there evidence of water entering anywhere else. The basement of the adjacent house is perfectly dry.

Based on overnight observation, where in one case more than five centimetres of water entered the furnace room (but nowhere else) between two inspections three hours apart, and the discovery of water seeping slowly from a crack in the corner of that room, the current theory is that as a result of the broken water pipe of January 2005 and the resulting disastrous water leak, sand and mud have migrated into the groundwater drainage system, which has partially, but not completely blocked an outlet to the cistern and pump system which evacuates it to the sewer (which is uphill from Fourmilab, wouldn't you know?). With normal precipitation, there remains enough flow to dispose of the water as fast as it arrives, but with the shock load of a big rainstorm triggering the melting of several months' accumulated snow, water backs up in the system, rising to the level of the crack (and potentially other entry points around utility pipes positioned well above the usual level of ground water), and entering the building. Unless you happen to observe it at the moment when the backed-up water is above the crack, all you see is a subtle trickle or drip.

We patched the crack with masonry filler as a (literal) stop-gap measure. If the patch holds, this should buy another 25 cm the water can rise before entering around the utility pipes. Early next week comes the digging out of the muck in the ground water conduits and live testing with water pumped into them.

If you're expecting something from me, don't expect it soon, and bear in mind that it may have been cleared entirely from mine due to limited stack depth in interrupt processing. Fortunately, all of the servers and communication gear associated with the site are at least three metres above the area which flooded, and the electrical substation has, so far, remained dry.

Posted at February 17, 2006 18:13