Flappers vs. siphons
Backflow prevention is what one calls a system for keeping a pump from turning the wrong way because of liquid flowing backwards. Backflow prevention became a huge issue during Katrina. When Jefferson Parish president Aaron Broussard made the career-killing decision to send his pump operators 100 miles out of town as Katrina approached - he didn't deem them essential personnel - he also prevented monitoring of the systems that kept Lake Pontchartrain's massive storm surge from backflowing through the pumps. When those systems (some of them gates, some of them a goofy compressed air system) failed, much of Jefferson Parish flooded. It didn't have to be that way. Not one pump in Jefferson Parish was damaged during Katrina, meaning much of the parish could have probably remained completely dry, if only the operators were there to run the pumps or check the backflow prevention systems.
Being a resident of New Orleans, I was curious about the backflow prevention measures for the new floodgate pumps. There are measures in place, but I wonder about their efficacy.
There are many ways to keep water from running backward through pumps. The easiest - and most reliable - way is to physically block the discharge pipe with something solid. This could be a sluice gate (a piece of metal that slides down across the pipe) or a flapper gate (a hinged metal door on the end of the pipe). The S&WB uses both on its drainage systems. On the three pump stations in New Orleans East that discharge directly into the Lake (10, 14, and 16), there are sluice gates. Along the Palmetto Canal in Gert Town and Hollygrove, there are numerous flapper gates to prevent backflow into the neighborhoods when water flows down the canal from station 1. Here's a picture of one of those gates:
Another way is a "siphon leg." This is an upsidedown-U stretch of pipe in the discharge. The idea is that if the peak of the "U" is tall enough, the water can't get over and into the pump. If that happened, a siphon (that is, a vacuum) would develop and water would flow backwards of its own accord. In fact, when the pumps are running normally, they take advantage of that siphon to keep water flowing away from the pump. The S&WB also uses siphon legs, but only in combination with sluice gates.
Just in case the sluice gates fail, and a pump breaks or is down for service, and a backwards flowing siphon does develop, there is a device called a "siphon breaker" at the peak of the siphon leg. The siphon breaker detects backward flow and opens a valve that introduces air into the leg, breaking the siphon, or vacuum.
What is notable about the siphon leg idea is that - by itself - there is no positive physical closure. If the pump stops and the siphon breaker fails there is nothing to stop the backflow. A gate closure is the nearest thing to a guarantee against backflow, as opposed to a siphon leg. That's why the S&WB installed sluice gates at stations such as station 14, shown below.
The pumps are the four gray things in the middle. In this picture, the water flows from the inlet canal at the right (just out of frame), up through the pumps and over the siphon leg horseshoes into an underground culvert. This culvert goes under Hayne Blvd. and the Lake levee to the left and out to Lake Pontchartrain. The sluice gates are mounted in the levee.
Note that the sluice gates, in order to be effective, must be present. The S&WB might want to look into a slight defect in that idea at station 16:
This is the outlet of the pumps at station 16 where it flows into the lake. There are no gates. They appear to have been stolen, removed, or otherwise misplaced. I hope the siphon breakers are in good working order at that station.
At the new floodgates
The Corps' new floodgate pumps use siphon legs, but no flapper gates. Here's the original drawing sent to pump vendors when the pumps were out for bid (jpeg version, click for bigger):
Here's a pdf of the same drawing. The pdf is about one tenth the size, if you want to save it to your harddrive
And here's a picture of the final installation:
This picture shows the siphon legs for three pumps at the 17th Street Canal. The blue devices on top of the siphon legs are the siphon breakers. They are made by a company called Wm. P. Wilson & Sons. Three other links on that page give technical information. The breakers at 17th Street are 12" versions. Note that the siphon breakers appear to be undersized, if a flow equivalent to the pumps' capacities (about 250 cubic feet per second) were to backflow. This is because the maximum flow capacity for a 12" Wilson siphon breaker is apparently 225 cfs, as seen here. Could this lead to another backflow problem in case of pump failures at the outfall canals? Yes. And in light of recent troubles with the pumps, the possibility is particularly distinct.
The Corps considered flappers, but removed them from the design upon the protest of the S&WB. Here is what the discharges of the floodgate pumps look like now:
Apparently, at some time in the past, the S&WB suffered a failure in one of those Palmetto Canal flapper gates I mentioned above. Something kept the gate open when water was flowing down the canal, and a section of Hollygrove flooded.
However, there is a fundamental difference in the situation that occurred on the Palmetto Canal and that which might occur at the floodgate pumps. Along the Palmetto Canal, the flapper gates were being pushed open by water flowing relatively slowly by gravity. Also, those gates are only 12" in diameter and are much more easily forced open by sticks, trash, and other debris. At the floodgates, the pumps would be (hopefully) running the entire time during the threat of backflow, i.e. during storm surge. Also, the flappers would be 108" in diameter, which is huge, heavy, and pretty much impossible to force open except by the flow of the water. They would be used much less and inspected far more often than the S&WB gates (simply because there are many more S&WB flappers), including right before they were used.
Note that the reason the S&WB has siphon legs on stations such as those in New Orleans East is not for backflow protection, though that is a side benefit of their design. The main reason for the siphon legs was to move the electrical equipment on the motors above the level of a major flood. They still wisely rely on the sluice gates for backflow protection.
Another disadvantage of siphon legs is that pumps must devote lots of extra energy to get started. This is because the pump must push the water up and over the leg. Without the siphon leg, the pump could simply push water through a relatively flat pipe to the outlet. It's equivalent to the difference between rolling a barbell across the floor as opposed to picking it up and carrying it across the room. In fact, at the floodgates, pumps must be started in pairs because of all the energy required to establish a siphon in the discharge pipe, as this excerpt buried on page 81 of the Floodgates Operating Manual describes (for reference, the "mainfold pipe" is the 108" pipe into which the 60" discharges of the pumps flow):
"Note that at least 2 pumps must run to establish a siphon in the manifold pipe and therefore the grouping of the 3 pumps in 2 stages was selected for pumps at the 17th St. Canal and London Ave. Canal. At the Orleans Ave. Canal the pumps are grouped with 3 pumps in the first stage and 2 pumps in the second stage."
What happens if one of the pumps in the second stage at Orleans Ave can't start? Well, that means that both the pumps in that pair can't start.
By the way, this is most likely why the Corps tested two pumps at Orleans Ave last week and three at London Ave., rather than doing individual tests of the pumps. It's because the individual pumps can't really be started very well by themselves.
One more question I have is this: What prevents backflow if the gates are lowered and the pumps aren't running? I mean, according to the Corps, it's very possible to have a big storm surge, but not a lot of rainfall. That's what they keep telling us in order to justify their wrongheaded decision not to place enough pumps at the floodgates. Well, if you take that argument to its logical conclusion, what if the storm surge precedes the rainfall by quite a while and they close the gates, but there's no need to run the pumps? Then the entire city is depending on those little siphon breakers to guard against a levee breach? That's pretty risky!
Flapper gates do work
As a postscript, it's not like there isn't precedent for installing what I'm describing, All one has to do is visit St. Bernard Parish - specifically the E.J. Gore pumping station:
E.J Gore uses pumps of the same manufacturer and type as the new ones at the floodgates. They are MWI hydraulically-powered axial flow pumps. These days, MWI calls them Hydraflo. Here's what the 60" Hydraflo's looked like this past July before they were put in place at the floodgates:
This was just off Marconi Drive near the Orleans Ave. floodgate site. The 60" (that's 5 feet!) I refer to is the diameter across the pump. The pump on the right is sitting with its bottom to the right of the photo. The pumps are turned up on end when they are installed at the gates.
The E.J. Gore pumps are the same as those at the floodgates, just a bit smaller. Actually they're 42" diameter pumps, and there's six of them:
The sixth pump is just out of frame to the left. They're all submerged in this picture.
What's interesting about the E.J. Gore station is that these MWI pumps have been in service for over 20 years without complaint. But there's an even more interesting part of the station. Look what's on the end of the pump discharge pipes:
Flapper gates! There are no siphon legs at E.J. Gore. I ask you, when considering your safety from backflow problems at the floodgates, would you feel more confident in what would undoubtedly be sturdier versions of the above (the pipe diameter in the picture is 42" - the pipe diameter for the manifold discharge at the gates is 108"), or 12" siphon breakers that - according to the manufacturer's own literature - appear to be installed outside their suggested operating range? Keep in mind that the consequences of backflow into this system are even more dire that those experienced by Jefferson Parish during Katrina, especially when one considers that:
a) The automated control and monitoring systems are not ready this storm season, meaning there is no way to remotely monitor or operate the pumps. According to the original bid solicitation, those systems were supposed to be installed 48 days after award of the contract, which was on April 14th. Obviously, the contractor, Prime Controls, is way behind the original schedule.
b) The Corps will pull its personnel out of the S&WB pump stations for anything over a category 3 storm, leaving the S&WB to fend for themselves. From page 9 of the Floodgates Operating Manual:
"Based on the predicted storm data from the EOC and the construction status at each structure, the Chief of Operations will determine if personnel will remain at the Outfall Canal Control Houses, be relocated to local safe locations or evacuated to remote safe locations. Until the control houses are complete, the Outfall Canal Teams shall report to the MVN reservation for a category 1 or 2 storm. For a category 3 or 4 storm, they will be stationed at the LADOTD building on Lakeshore Dr. For a category 5 storm they shall be evacuated to a remote location. Canal Captains shall be stationed at the S&WB Pump Stations for up to a Category 3 Storm [emphasis mine]. For a storm above Cat. 3, they will follow the plan for the Operations Teams."For reference, the Control Houses are complete at London Ave. and Orleans Ave., but not at 17th Street.
If there is backflow through the pumps, the possibilties for craziness are legion. It would seem to be theoretically possible to get water flowing down the outfall canals once again! I wish the Corps had just gone with the flapper gates.