![]()
The system was designed by the U.S. Army Corps of Engineers Savannah District Office (CESAS) during FY93 in cooperation with DPW Fort Jackson and CERL.
Design Goal
The design goal of the chilled water storage (CWS) system for the Central Energy Plant (CEP) No. 2 at Fort Jackson was to shift operation of the four 120-ton chillers, from the summer on-peak hours (1300-2100, Monday-Friday excluding holidays) to off-peak hours. Recall that Fort Jackson has a master meter for measuring the installation-wide electrical demand for billing purposes. Figure 1 shows the hourly demand profile for the peak day in 1989.
Based on the new chiller rating of 0.64 kW/ton, turning off the four chillers (1200 ton each) at CEP No. 2 would reduce the electrical demand (P) by:
P = 4*1200 (ton)*0.64 (kW/ton) = 3072 (kW)
According to the hourly demand profile shown in Figure 1, reduction of 3000 kW in demand could be achieved by turning off the four chillers from 1300-1800 hours. Therefore, the goal of the system design was to shift the electrical demand of 3000 kW from on-peak to off-peak periods.
Central Energy Plant No. 2 Operating Conditions
CEP No. 2 has four chillers, each rated at a 1200-ton capacity. CEP No. 2 serves more than half of the major buildings at Fort Jackson. Main supply lines branch from the plant in two zones. The total flow rate for CEP No. 2 is 8300 gal/ minute (gpm) for the two zones, Zone 1 and Zone 2. The normal differential temperature of the chilled water is 12 °F with supply water temperature of 42 °F, and return water temperature of 54 °F. At night, the plant has a minimum of 2000 ton of cooling capacity to charge the tank.
Figure 1. Hourly demand (in MW) on the peak day in 1989 (12 July 89).
Recall that CEP No. 2 serves more than half of the major buildings at Fort Jackson, which requires cooling round the clock.
Tank Sizing
The storage capacity of the tank should be large enough to store enough cooling to meet the cooling demands of CEP No. 2 for a selected period of a day. Based on the operational data from CEP No. 2, the integrated cooling load from 1300 to 1800 hours was estimated to be 16,000 ton-hr. With a 5 percent safety factor, the design cooling capacity (Q) of the storage tank was determined to be:
Q = 16,800 ton-hr
The storage volume (V) of the tank is determined by the cooling capacity (Q), differential temperature between the supply and the return water (dT), and the figure-of-merit (FOM) of the storage tank. For a storage tank with well designed diffuser system, the FOM is recommended to be 0.9. (ASHRAE 1993). The temper-ature differential for CEP No. 2 is 12 °F. Based on the these data, the volume of the storage tank was determined by:
V = Q/[cp *dT*FOM]
= 16800 (ton-hr) *12000 (Btu/ton-hr)/[1(Btu/lbm _F) *12 (_F)*0.9 * 8.36 (lbm/gal)]
= 2.232M (gal)
The volume of the storage tank was determined to be 2.25M gal.
Tank and Diffuser Design
Due to the multiple competing options in the market (such as concrete or steel tank, and linear or radial diffuser), the CESAS bid specifications prescribed the functional requirements for the tank. The functional requirements included the storage volume, the storage cooling capacity, and the design of diffuser system as well as the other requirements such as aesthetic requirements from Fort Jackson. A particular aesthetic requirement was the height of the tank, which was not to exceed that of the nearby facility. The maximum height of the water column in the tank was limited to 40 ft. The bidders were requested to provide the details of the design of their choices, and the designs were subjected to the approval from the design team of the Savannah District Office, Fort Jackson, and CERL. The important characteristics of the approved design follow.
Tank Configuration. Tanks with low surface-to-volume ratios have a lower degree of thermal loss and have a lower cost per ton-hour of stored cooling construction cost. Therefore, flat-bottomed vertical cylinders are favored. Concrete tanks with height-to-diameter ratios between 0.25 and 0.33 represent a good compromise between a low-cost short tank and a tall tank that provides the best thermal stratification (ASHRAE 1993). Other factors must also be considered when determining tank dimensions, such as required flow rates and dimensions of the diffuser, and site conditions. The allowable bearing capacity of the soil as well as special architectural concerns should be taken into consideration. One particular criterion for the tank design at Fort Jackson was that the height of the tank not exceed the height of the tallest structure in its vicinity. A 40-ft tank water level, with a maximum tank height at 44 ft, was chosen. The resulting diameter was calculated as:
H = 40 ft
D = 2*(V/_H)(1/2)
= 2*((2,25M gal)/(7.48 gal/cu ft)(_)(40 ft))1/2 = 98 ft
An aboveground cylindrical concrete tank was chosen with a 40-ft water level and 98-ft diameter. The resulting tank height-to-diameter ratio was 0.41, which favors thermal stratification in the tank. The tank was built on a reinforced concrete ring wall. A reinforcing steel rod skeleton was constructed and an encased inner steel shell was attached to the skeleton. Shotcrete was applied and allowed to cure and insulation was added to the walls. A synthetic stucco-covered exterior was added to the exterior of the tank for aesthetic reasons. Table 1 summarizes tank design characteristics. Figure 2 shows a diagram of the tank, including its elevation. Figure 3 shows the Tank Plan. An earlier paper reported to the 1995 USACE Electrical and Mechanical Engineering Training Conference (Burch 1995) discusses the tank's design in detail.
Internal Diffuser Design. A chilled water storage tank needs diffusers to introduce water into the tank without creating disturbances in the fluid that could result in the deterioration of the thermocline. During charging the tank, a gravity current of cool, dense water is produced by the lower diffuser near the tank floor, and is spread horizontally. Similarly, a gravity current of less dense warm water is produced near the top of the tank by the upper diffuser during discharge. Octagonal diffusers, formed from eight straight sections of pipe connected with 135-degree elbows, have proved successful in the past for creation and maintenance of the thermocline in the tank. Octagonal diffusers with both the lower and upper array consisting of four rings were chosen to ensure proper stratification. The upper and lower diffusers were identical in shape. A maximum of 20 psid pressure loss from inlet flange to outlet flange was specified, as well as design flow rates of 4,000 gpm (charging) and 8,000 gpm (discharging). Due to the competing technologies and builders in the market, the CESAS design left the actual design of the internal diffuser to the contractor to be selected through an open bidding. However, the perform-ance of the diffuser was prescribed to meet the industry recommendation of the maximum inlet Reynolds Number to be less than 2000, and the Froude number less than 2 (ASHRAE 1993). The final design of the diffuser by the successful bidder was a quadruple octagonal diffuser system with the total linear diffuser length of 851 ft (Figure 4). Note that Figure 4 is intended to convey the general configuration of the diffuser system, not the fine details of each segment.
Cooling capacity |
16,800 ton-hour |
Size |
2.25M gal |
Mean diameter |
98 ft |
Water level height |
40 ft |
Height-to-diameter ratio |
0.41 |
Plan area |
7,543 sq ft |
Vertical core wall thickness |
Tapers from 7 ¼ in. at bottom to 3 ½ in. at top including 1 in. cover over steel shell diaphragm. |
Vertical wall material |
Prestressed composite wall (steel shell/shotcrete) with 5 in. thick rigid Styrofoam insulation glued to concrete tank and finished vee rib outer sheeting; brick outer shell for bottom 7 ft-8 in. |
Dome shell |
3-in. thick concrete with expanded polystyrene insulation |
Floor |
5-in. concrete with painted outer surface |
Table 2 lists detailed characteristics of the installed diffuser system.
An underground 24-in. chilled water supply pipe and 24-in. chilled water return pipe connected the storage tank to CEP No. 2. The flow branched into two separate 20-in. PVC pipes and was carried to the first octagonal ring. All octa-gonal ring diffusers were 14-in. PVC. Total diffuser length, which represented the sum of the four octagonal rings, was 851.0 ft. The sections of pipe had slots cut into them, through which the flow was diffused into the tank. The area of each slot was set such that the sum of the slot areas in each diffuser pipe equaled the cross-sectional diffuser pipe area. The calculation of an effective diffuser length based on twice the total diffuser length was necessary to account for the fact that the water is diffused into the tank in both the radial inward and radial outward directions. Total flow rate was based on 125 percent of the design discharge rate and was calculated as:
(8,000 gpm)(1.25)/((60 sec/min)(7.48 gal/cu ft)) = 22.3 cu ft/sec
Since the diffuser design included 32 sections of pipe, each diffuser pipe flow rate was calculated to be:
(22.3 cu ft/sec)/32 = 0.7 cu ft/sec
Mixing in the tank is also influenced by the inlet flow rate per unit length of diffuser, expressed by the Reynolds number. A maximum Reynolds number of 2,000 was specified for the design of the diffuser. The Reynolds number is defined as the flow rate per foot of diffuser length divided by the kinematic viscosity of inlet water. The total diffusion rate was:
22.3 cu ft/sec/1,702 ft = 0.01310 sq ft/s
Table 2. Diffuser characteristics.
Design inlet temperature during discharge cycle |
54 _F |
Design outlet temperature during discharge cycle |
42 _F |
Design inlet temperature during charge cycle |
41 _F |
Reynolds number |
811 (2,000 design) |
Design Froude number |
0.5 |
Each diffuser pipe flow rate |
0.7 cu ft/sec |
Number of slots in each diffuser pipe |
34 |
Flow rate of each slot |
0.0206 cu ft/sec |
Cross-sectional diffuser area |
1.0 sq ft |
Slot area |
0.0294 sq ft |
Slot inlet/exit velocity |
0.70 ft/sec |
Slot width |
3/8 in. |
Slot length |
11.4 in. |
Length of each diffuser pipe in octagon #1 |
14.0 ft |
Length of each diffuser pipe in octagon #2 |
24.2 ft |
Length of each diffuser pipe in octagon #3 |
31.2 ft |
Length of each diffuser pipe in octagon #4 |
37.0 ft |
The Reynolds number was calculated as:
Re = (0.01310 sq ft/sec)/(0.00001615 sq ft/sec) = 811
This Reynolds number value was less than the design maximum allowable (2000) recommended in the industry standard design guide (ASHRAE 1993).
The Froude number is defined as the dimensionless ratio of the inertia force to the buoyancy force acting on a fluid. Gravity currents, which are necessary for the proper performance of the tank, will form for Froude numbers less than 1 with limited mixing. The Froude number criterion is used to determine the required inlet height of the diffuser. For a diffuser close to the bottom of the tank, the inlet height is defined as the distance from the tank floor to the top of the diffuser inlet opening. For a chilled water tank, the Froude number is defined as:
Fri = q/[gh3(_i-_a)/_a]1/2 Eq. 1
where:
q = volume flow rate per unit diffuser length
g = acceleration of gravity
h = minimum inlet opening height
_I = density of inlet water
_a = density of ambient water
and
q = Q/L Eq. 2
where:
Q = maximum flow rate
L = effective diffuser length.
A Froude number of 0.5 was selected for design of the inlet opening height. Other values necessary for the calculation included:
q = 0.01310 sq ft/sec
g = 32.17 ft/sec2
_i = density of inlet water
_a = density of ambient water.
Solving for the inlet height yielded a value of 0.35 ft (which, in the final design, was set at 4 in.). The bottom of the lower diffusers were placed 2 in. above the tank floor. The top of the upper diffusers were placed 4 in. from the free surface in the original design.
System Schematics
Four primary chilled water pumps (PCWP) serve the chillers. Two regenerative turbine pumps (RTP) serve the chilled water tank and one heat exchanger pump (HEP) serves the heat exchanger for free cooling application. Also included are four system chilled water pumps (SCHP). Each of the pumps has the characteristics described in Table 3. Two-way, two-position, normally open electric solenoid actuated pneumatic valves are located at each of the four PCHPs. Temperature sensors were mounted inside the storage tank to measure the vertical temperature profile and thermocline thickness in the tank. Figure 5 shows a schematic of the chilled water storage system.
Construction of the system was accomplished in two phases. In the phase 1, the tank with internal diffuser system was built in 1994 (Figure 6). Figures 2 (p 13) and 3 (p 14) show the plan and elevation of the tank with the internal diffuser system installed inside the tank. Figure 4 shows the layout of the diffuser segments. Note that the top and the bottom diffusers are a mirror image.
Phase 2 consisted of adjusting the piping inside the energy plant (CEP No. 2) and the pipe connection from the storage tank to CEP No. 2 during the off-cooling season of 1995-96. The phase 2 work was completed in March 1996. During the commissioning of the tank to the cooling loop of CEP No. 2, a major breakage of the upper diffuser assembly inside the tank was detected. Figure 7 shows a typical breakage of the diffuser. The tank was drained, the cause of failure was investigated, and the upper diffuser assembly was repaired for a successful system commissioning on 20 May 1996.
Table 3. Chilled water system pump characteristics.
RTP |
PCHP |
SCHP |
HEP | |
GPM |
2,000 |
2,060 |
2,600 |
2,000 |
Total head (feet) |
70 |
25 |
175 |
35 |
Max horsepower |
80 |
20 |
150 |
20 |
RPM |
1,750 |
1,170 |
1,750 |
1,170 |

Figure 6. 2.25M gal chilled water storage tank at Fort Jackson, SC.
Figure 7.
Breakage of upper diffuser assembly.
Breakage and Repair of Upper Diffuser Assembly
Note that the upper distribution diffuser system (shown in Figure 2, "Elevation of Tank") is hanging from the ceiling with 3/8-in. stainless steel threaded rods fixed to the dome roof. About 26 breakage points in the upper diffuser system, including diffuser and riser (feeder line to diffuser), were noticed (see Figure 7). The postulated causes of failure and repairs made are:
1. Buoyancy on the diffuser due to air pocket. When the tank was initially filled with water, the tank was not connected to CEP No. 2. Water was introduced through the opening at the ceiling. The two 24-ft main transfer lines (to and from the tank, shown in Figure 2) remained closed by isolation valves. Water was introduced into the diffuser assembly through the slots into the closed pipe space. The lower diffuser assembly is anchored to the concrete floor with aluminum mounting pads between the diffuser and floor. The pad and anchor holds the lower diffuser assembly securely against any potential buoyancy forces. On the other hand, the upper diffuser is secured by the 3/8-in. rods, which cannot provide resistance to compression induced by potential buoyancy force due to air pocket inside the upper diffuser assembly. The solution was to install a bypass line across the two 24-in. main transfer lines. A short 11-in. piping with a shut-off valve in the middle was installed between the two main transfer lines just outside the tank. When filling the tank, the bypass line will equalize the rising water level between the inside and outside of the diffuser system, thereby eliminating any potential buoyancy effects.
2. Valve actuating speed. Line-sized butterfly control valves are installed at the outlet of recovery turbines in CEP No. 2. The opening speed of these valves was so fast that it could induce water hammer effects along the line, including the diffusers inside the tank. The solution was to slow the valve opening and closing speed to 60 seconds for full opening and closing the valves.
3. Loose connection of main transfer line to upper diffuser. The flange bolts connecting the 24-in. steel main transfer line from CEP No. 2 to the 24-in. PVC line to the upper diffuser assembly (marked with an asterisk [*] in Figure 2) within the tank were missing nuts underneath the flange. The loose connection would have generated a significant flow-induced vibration of the upper diffuser structure when a full charge flowrate was introduced to the tank. The flange nuts were installed and tightened for a secure connection of the 24-in. main transfer line for the upper diffuser.
4. Leveling of Upper Diffuser. The broken parts of upper diffuser were fixed and the tank was fully charged with city water. One major concern was the levelness of upper diffuser segments. A DPW engineer entered the tank and measured the elevation of high spots along the diffuser segments. The maximum elevation differential at the highest spot was measured to be 5 in. The original design water depth between the top surface of water and the highest point in the diffuser was 8 in. With the unevenness of up to 5 in., the operating water depth would be reduced down to 3-in. at the highest spot. In case of potential rapid loss of water in the system, e.g., a rupture in the distribution line, the 3-in. margin was deemed too shallow to prevent potential exposure of diffuser slots to atmosphere. Exposure of slots to open air will result in the introduction of air into the circulation system. The solution was to raise the operating tank water level by 7 in. by extending the overflow level from 40 ft to 40 ft 7 in. The tank builder was consulted for the safety of heightened level of water inside the tank, and confirmed the safety of the tank with the increased level of 7 in.
Commissioning of System
The tank was fully filled with city water. Chillers in CEP No. 2 completed charging the tank with chilled water during the weekend of 18 May 1996. The tank was fully charged by early morning 20 May (Monday). The temperature profile inside the tank ranged from 40 °F at the bottom to 43 °F at the top.
The ambient temperature in Columbia, SC on 20 May reached 99 °F. By noon, all four chillers (1200 ton each) in the Energy Plant No. 2 were running to provide cooling for Fort Jackson. Starting from 1222 (20 May 1996), all four chillers were shut down: No. 1 chiller at 1222, No. 2 at 1252, No. 3 at 1307, and No. 4 at 1320. Note that the utility on-peak hours for Fort Jackson are between 1300 and 2100. The chilled water in the tank met the entire cooling load during the peak hours. Chillers were brought back online starting at 1622 for No. 1, 1007 for No. 2, 1722 for No. 3, and 1807 for No. 4. This operation helped Fort Jackson keep its on-peak billing demand under 19,550 kW (Figure 8, "Hourly Load Profile of Fort Jackson, 20 May 1996"). On 20 May 1996, the electrical demand was peaking around 1100 at 23,000 kW. Without the shutdown of the four chillers, the demand should have increased to over 23,000 kW in the early afternoon hours. Therefore, the minimum amount of peak shaving by the storage tank is 3450 kW (the difference between 23000 kW and 19550 kW). Table 4 lists the thermal performance of the tank for the first complete cycle of charging and discharging. The table shows the temperature distribution inside the tank at a number of benchmark hours. Note that, for the first day of operation (20 May 1996), the tank was not fully discharged. Table 4 confirms the regenerating capability of the tank through the night of 20 May. By the morning of 21 May, the tank was fully recharged and ready to repeat the cooling cycle.
Table 4. Temperature distributions inside the tank.
Sensor # |
Date/Time 20 May 1996 (1710 EDT) oC/oF |
Date/Time 21 May 1996 (0830 EDT) oC/ oF |
Date/Time 22 May 1996 (0830 EDT) oC/ oF |
20 (top) |
13.8/56.9 |
6.7/44.0 |
7.3/45.2 |
19 |
13.5/56.3 |
6.7/44.0 |
7.2/44.9 |
18 |
13.2/55.8 |
6.6/43.8 |
7.0/44.6 |
17 |
12.8/55.0 |
6.5/43.7 |
7.044.6 |
16 |
11.8/53.3 |
5.9/42.7 |
6.4/43.6 |
15 |
11.1/51.9 |
5.9/42.6 |
6.1/42.9 |
14 |
10.6/51.1 |
5.4/41.7 |
5.8/42.5 |
13 |
10.2/50.4 |
5.3/41.5 |
5.8/42.4 |
12 |
9.9/49.9 |
5.3/41.5 |
5.7/42.3 |
11 |
5.5/41.9 |
4.9/40.9 |
5.4/41.7 |
10 |
4.8/40.7 |
5.3/41.5 |
5.7/42.3 |
9 |
4.3/39.8 |
4.8/40.6 |
5.2/41.3 |
8 |
4.2/39.6 |
4.7/40.4 |
4.9/40.9 |
7 |
data missing | ||
6 |
data missing | ||
5 |
data missing | ||
4 |
4.8/40.7 |
4.8/40.6 |
5.5/41.9 |
3 |
4.9/40.9 |
4.8/40.7 |
5.7/42.2 |
2 |
4.9/40.9 |
4.7/40.4 |
5.5/41.9 |
1 (bottom) |
data missing | ||