The Ballast Tanks

The Ballast Tanks

arge openings expected above the ballast tank bulkheads were confirmed when excavation reached the ends of the cabin.

The openings, described in at least one report as nine inches high, are probably large enough to permit some access to the tanks from the cabin for cleaning or for maintenance. Although the open-topped tanks seem dangerous to us, and were responsible for at least one sinking, we can view the Hunley design as a technical improvement over the Revolutionary War Turtle, since that pioneering submarine used the bottom of the cabin as a tank, giving the operator wet feet.

The ballast tanks were filled by opening stopcocks located in the forward and aft hatch sections of the cabin (red in the graphics below). One would be operated by the captain and one by the crewman in the most aft crank station. These would be closed when the tanks contained sufficient water, perhaps upon a shouted command from the captain or after a predetermined count.

The tanks were emptied using the pumps located in the hatch sections (green in the graphics). The aft pump handle is situated within reach of the aft crewman. Removal of the bellows exposed the forward ballast tank pump handle running across the starboard side of the first crew station, revealing that the captain did not operate this pump, as previously believed. At least for the final mission the first crewman, identified as Becker, actually had two stations. For cranking he occupied the second station. He could operate the air bellows or the forward pump from the first station or perhaps he could reach the handle from the second position. The tanks have the same stiffener rings at about 2-foot intervals as the cabin. In addition to strengthening the tapered hull against compression, these rings would also have acted as baffles to limit ballast water sloshing. The Pumps

In 2004, the Friends revealed that the pump plumbing is much more complex than previously thought. The forward and aft tanks are connected by a pipe running under the bench and a complicated system of valves controls at each pump. Early Published photos of these areas are not very helpful because of the heavy concretion, but we can speculate.

Pump valving schematic

This analysis addresses only one pump, but we can assume the same arrangement would be used for both.

All valves would be accessible in the vicinity of the pump by one person.
There is an historical report that one of the pumps became clogged and had to be opened and cleared. If this is true the pump would need valves to isolate it from its input and output lines. That means 2 valves, one on the input line and one on the output.
Considering the Friends report that the pump could input from both tanks and from the cabin floor, it would probably be necessary to isolate the three sources. It would definitely be necessary to isolate the floor from the tank and it would be prudent to isolate the two tanks. This translates to 3 input valves.
If, as mentioned in the report, water could be pumped from one tank to the other two output lines are necessary, one to the tank and one to the sea. Both of these would need to be valved.

So I speculate 5 valves per pump if active pumping from tank to tank is possible and 4 per pump if not.

Early speculative reconstruction of the forward pump

The early reconstruction at left pictures the forward pump and its plumbing, as interpreted from early published photos. The fifth line and valve, the green valve in the schematic above - pump output to the forward tank - are more problematic and not shown. The handle and the upper part of the pump appear to be bolted to the nearby stiffener ring, partly ghosted in the graphic. The lower pump body and plumbing were completely speculative. See "More General Speculation" below for later thoughts.

My reconstruction shows handles on each of the valves, but some reports indicate that some are not attached.

The Summer 2008 Blue Light included two very nice photos of the aft pump during the later stages of conservation. Some parts of the assembly were obscured or not shown in the photos, but these were still the best photos yet published.

The Friends released images of 3D scans of the aft pump in October 2008. (My interpretation is at right.) The images show almost the entire assembly - the exit valve has been removed from the side flange and a cleanout cover, still attached in the Blue Light photos, is missing (but included in my image). Oddly, the upper part of the pump has been rotated about 90 degrees so that the exit valve would attach on the side of the pump rather than the front as shown correctly in my image. There is a flange at the bottom of the pump barrel attached to a similar but differently sized flange at the top of the lower zigzag piece. The pump assembly must have been detached here during conservation and turned for the Friends' image. I wonder if this reflects an original configuration before the pump was installed in the Hunley. Or perhaps the top and bottom parts come from different places. See my general speculation below.

Aft pump - from FotH renderings of 3-D scan

Forward pump operation (early reconstruction)

The pumps appear to be a cross head types as first identified by Kevin Wilcomb. The forward pumping mechanism, pictured in operation at left, is fairly clear in published photos.
As of October 2008, there are no published images of the cleaned-up forward pump. I've used the aft pump images to improve my forward pump reconstruction images.

One clear difference between the pumps is the much lower location of the aft pump handle. As depicted at right, I used features visible in published images of the aft pump to show how a similar mechanism might have worked. A long rod that I once thought was a valve handle serves as a link between the low-mounted handle and the pump piston rod, permitting the major parts of the two pumps to be identical. (Note that the animation view is looking toward the portside, opposite that in the animation above.) The animations use older, and different, less accurate pump reconstructions.

More General Speculation

Removal of both pumps in 2008 provided new images and considerable new information. The pumps appear to have many similarities and a few differences. The graphic below shows top-down views.

Forward pump speculative resconstruction

The forward pump is located on the starboard side and the aft pump is located on the port side.

Aft pump speculative resconstruction

The outlet pipe from each pump runs forward and then turns sharply to the near cabin wall where it is attached with a square plate. The forward outlet exits on the starboard side of the hull, which in this section is tapering inward; the aft outlet exits on the port side at nearly the same vertical location, but the the hull is widening instead of narrowing. Both outlet pipes are valved near the pump. The valve handle for the aft pump valve is on top of the outlet assembly while the forward valve handle is on the bottom, possibly to keep the outlet pipe top open as a support the commander's seat. The two pumps have similar upper sections joined by a flange that zigzags to a lower section that attaches to the inlet plumbing, as shown in the graphics below. On the bottom of the lower section there is a round or oval plate with four bolts for attachment to the hull. The forward pump was attached to the hull by a bracket at the top and the aft pump likely was as well.

The main parts of the two pump assemblies appear reversed and very nearly symmetrical. I believe the basic pumps are identical with differences only to accommodate installation in slightly different locations. The historical literature says that the Hunley was constructed from locomotive parts. There is some discussion of this on other parts of these pages. Kevin Wilcomb has provided me a photo of an early 1800s locomotive with a water pump similar to those on the Hunley. Salvage from earlier construction, whether railroad or maritime, is consistent with manufacturing resource conditions in the Confederacy, as described by Tom Chaffin in The H.L. Hunley: The Secret Hope of the Confederacy.

The pumps may have come from two sides of a single locomotive or the pump bodies may have been split at the lower joining plates and reversed to provide room for the valve assembly in each location. As mentioned above, the top and bottom parts may have been independent originally.
The graphic at left shows the aft pump. I've added the full handle and a speculative reconstruction of the outlet assembly, shown in white. The graphic below shows a reconstruction of the forward pump based on the Friends' aft pump renderings and many published photos.

Displacements and Capacities

     The National Park Service Site Assessment Report calculates the Hunley's displacement two ways, one a simple estimate based on a 4-foot-diameter cylinder and one using actual hull lines. The credibility of the hull lines calculation is suspect since both methods yield the same number - the one based on the simple cylinder. The goal of the Site Assessment estimate was to bound the weight of the Hunley to assure the lifting method chosen could be safely and successfully completed.
     I've undertaken my own analysis based more or less on my speculative dimensions. The analysis is sensitive to to a number of sometimes arbitrary parameter values but I believe it is reasonably accurate. The approach I took uses the likely hull volume, reasonable estimates of the internal space volumes, and historical data.
     I calculate total displacement volume for the Hunley at about 350 ft³ displacing about 22,600 lbs of sea water. I estimate the dry weight of the hull at about 19, 500 lbs, and with the crew and their gear aboard, about 20,800 lbs. This translates to a positive buoyancy of about 1800 lbs, requiring about 28 ft³ ballast tank capacity. My estimated tank capacity of greater than 45 ft³ meets this requirement with sufficient margin to avoid spillage over the bulkheads.

Alexander provides the Hunley's draft in his rough 1902 sketch, detailed at right. The lower arrow indicates the "waterline light", presumably with tanks empty.

Detail from Alexander's 1902 sketch

The upper indicates "waterline when ballasted to sink", presumably with tanks filled. Since Alexander shows the hatches above water, the Hunley must have been operated with positive buoyancy; the dive planes were required to submerge and keep the boat submerged. The difference between the two waterlines is less than a foot. Using this figure, I estimated the "sink" or heavy displacement based on the submerged portion of the hull as about 22,300 lbs, using 1500 lbs of ballast water. As dramatized in the animation below, this is only about half of the ballast tank capacity and well below the top of the bulkhead. The tops of the hatches that Alexander leaves exposed in his "sink" configuration present a positive buoyancy of about 300 lbs that could be overcome with an increase of 20% in the water ballast to the 28 ft³ mentioned above.

Analysis

Bill Akins has raised some basic questions about how the tanks would fill. If the cabin was sealed when the tanks were filled, a reasonable assumption, the air in the cabin would be compressed as it was displaced by water. When the external and internal pressures were the same, no more water could enter. The physics are simple and familiar to anyone who has placed an inverted glass in a tubful of water.

I calculate the air volume in the cabin with crew aboard and no water in the tanks at about 290 ft³. Using Alexander's nominal foot or so of increased depth, the 23 ft³ of ballast water compresses the air volume in the cabin to about 92% of its maximum volume, increasing the its pressure to about 16 psi. We can speculate about what this means but we don't have enough information to answer all the questions or solve the problem.
The following is a static analysis or more accurately, a set of discrete hydrostatic analyses based on the concept that pressure in a static fluid is dependent only on depth and the shape of the container is irrelevant. This is true, however, only when the fluid is a contiguous body, that is, when the fill outlet is under water in the tank. This is illustrated in the figure at right. When the cabin is sealed so air cannot enter or leave, ignoring any dynamic effects, filling stops when P₂, cabin pressure, equals P₅, the outside water pressure at the level the tank is filled to.
Determining characteristics when the water is flowing into the tank requires a more complex hydrodynamic analysis involving some or all of the four pressures noted:
P₁ - Ambient air P₂ - Cabin air P₃ - Fill inlet P₄ - Fill outlet P₅ - (no relationship with internal pressures)

The graph at right shows that with my assumed inlet pipes, shown in the plumbing speculation graphics near the top of this page, the aft fill outlet submerges but the forward one does not. I have not done the detailed analysis here, but I assume with the cabin otherwise sealed, air would escape via the forward fill pipe when the cabin pressure built up sufficiently.
The next graph shows how pressure changes as the tanks are filled to go from light to heavy configuration. The green curve shows the cabin pressure increase as the tanks are filled.

The gray curve shows the equivalent external pressure at the tank water surface level. If both tank fill outlets were submerged this would be the maximum pressure possible in a sealed cabin. Notice there the cross-over about halfway through submerging.

Pressure vs. depth in psi

The lines track from this point and are close enough that they may represent an error in my numbers. Nonetheless, the different speculative positions of fill outlets might result in different water levels in the two tanks. As I stated earlier, much of this analysis is based on speculation. I've made several assumptions about the plumbing that affect the results. There are several possible operational explanations and alternate configurations that can be considered.


o	The tanks could have been filled with the snorkel valve open, allowing air to escape and maintaining the standard ambient 14.7 psi.
o	The inter-tank valve could have been open during the filling operation.
If air was vented during the operation, the cabin air would not have been appreciably compressed in filling the tanks and that there would be a negative change when the tanks were emptied. This would result in difficulty in opening the hatches unless air were let in through the snorkel.
o	I have placed the forward inlet based on examination of several pictures of the forward bulkhead for the cabin. Since my first posting of this discussion the Friends have published a low resolution photo of the bulkhead from the tank side showing the position of the inlet and Brian Hicks has confirmed there is no pipe in the tank routing the inlet nearer the bottom.
o	Available views of the aft bulkhead area are extremely cluttered. I am confident of my valve placement but the routing of the plumbing to the bulkhead is speculative. I've placed the tank inlet to avoid the flywheel. Brian Hicks emphasizes the difficulty of distinguishing the mechanics in aft section but supports this placement based on his observations.
Based on the preceding analysis it seems to me the forward and aft fill outlets would work best at the same level, that is, the two red lines on the graph should be the same. Of course, the two tanks are connected and the valves could have been opened to allow the tanks to fill together. I will update the above discussion when and if more information becomes available. Interestingly, if the part of the above analysis that predicts a maximum tank level with positive buoyancy is correct, we may have a more acceptable explanation for Horace Hunley's fatal oversight in the second sinking. It may be that the tanks filled to only a certain level when the valves were opened and did not require close monitoring. Perhaps Hunley was distracted during what may have been several minutes before the sub, tanks full, began forward motion and active control with the plane. As the boat dove, the external pressure would have increased and the forward tank would have drawn more water....

What do you think?

E-mail me with comments