1、 Reza Ghias is the Director of Advanced Simulation Center (ASC) and Mike Kilkeary is an Associate Principal Engineer, both from Southland Industries located in Dulles, Virginia CFD Design and Validation of a Thermal Storage Tank System and Its Impact in a Design-Build Project Reza Ghias, DSc Michael
2、 Kilkeary, PE ASHRAE Member ASHRAE Member ABSTRACT The use of advanced engineering tools in HVAC industry has increased as the cost of computational fluid dynamics (CFD) has become more affordable for engineering firms. A thermal storage tank system utilized a series of vertical tanks to store chill
3、ed water and supplement chillers in the event of a power failure. Chilled water is routed from the storage tanks to critical equipment during chiller re-start, bridging the gap in time that the chillers are unable to provide set-point chilled water due to power failure and subsequent required time t
4、o re-start. The main challenge of the tank sizing was to optimize the tanks in order to maximize the output time of stored chilled water while minimizing the height and quantity of the tanks. The goal was achieved through an innovative design in which the tank diffusers were designed through a serie
5、s of simulations. Once installed, the system was tested rigorously to the same conditions that were previously simulated during design. The result of field-testing verified that the CFD model was accurate within two percent margin of error. Utilization of this process allowed the design-build team t
6、o save installation time, money, materials, and building square footage by reducing the number of required storage tanks to provide the required stored chilled water quantity. INTRODUCTION One of the many benefits of the Design-Build project delivery method is that the owner can define their needs a
7、nd allow the design-build contracting team the flexibility and the freedom to develop innovative solutions to meet their needs, not be constrained by rigid prescriptive design requirements, all wrapped into a competitive bid environment. Contractors enjoy the challenge of using design creativity to
8、meet the owners needs, reduce the cost of the project, and deliver the project faster than traditional design-bid-build delivery. The price of that freedom for contractors is the design risk. The Request for Proposal (RFP) response phase of a design-build project is generally a relatively short peri
9、od of time in which the contracting teams have to identify the key design challenges of the project, develop engineering solutions, and balance price and schedule ramifications of those solutions with the overall project constraints. The competitive bid environment encourages design teams to utilize
10、 innovative thinking, while the short RFP response period limits the amount of technical development the design engineering team can incorporate into the overall solution (Zhai 2006). This juxtaposition of requirements for the design team promotes them to utilize non-traditional methods for verifica
11、tion and validation of their design. CFD analysis, once considered too time consuming and expensive for conventional architectural/engineering building design, is now a design tool that is frequently used to predict results and remove ambiguities of the design process. Developing reliable CFD models
12、 for industrial applications and utilizing high performance computing (HPC) machines provide quick turnaround alternatives for the design team. The team can use these results to make design decisions, influence alternatives with respect to cost and scheduling, and create presentation graphics and vi
13、deos that convey the design solution to the owner with more than a “just take our word for it” response. For the project referenced in this technical paper, the task was to design and build a chilled water thermal storage system that would provide fifteen (15) minutes of chilled water in the event o
14、f a power loss and subsequent loss of active chillers. Fifteen minutes was required in this data center application in order to bridge the gap in time that the chillers are unable to provide set-point chilled water due to power failure and subsequent required time for those chillers to re-start and
15、energize to full capacity. An optimal design would maximize the output time of stored chilled water while minimizing the height and quantity of the tanks required. The solution would need to be robust enough to maintain a tight tolerance on chilled water supply temperature and distribution time, but
16、 simple enough to maintain competitive cost control. Many parameters can affect the performance of the storage tanks, and extensive experimental and numerical studies have analyzed these parameters (Karim 2009, Ghadar 1989, Bahnfleth 2003, Yoo 1986, Zurigat 1991, Stewart 1992). Figure 1(a) shows a t
17、hermal storage tank with charge and recharge diffusers at the top and bottom of the tank to distribute and recollect the water in the tank uniformly. The uniform flow distribution is more crucial in shorter tanks in which flow turbulences can deteriorate the thermocline that are formed due to buoyan
18、cy effect. Hudson et al. (1979) found that a ratio of half between the total opening area in the diffuser branches to the cross sectional area of the corresponding branch pipe can help the flow uniformity. In other words, the total slot area at each branch should be half of the heara of the pipe tha
19、t slots are created on. Figure 1 (a) Schematic of the thermal tanks to provide cool water at 60 F for at least 15 minutes. (b) Schematic of the double-ring diffuser with different slot sizes, which are highlighted on each branch of the inner and outer loops. The low velocity at inlet slots is anothe
20、r key factor to keep buoyancy force dominant enough to form the thermocline effectively. Dorgan and Elleson (Dorgan 1994) investigated the swirling in the tank as a result of non-uniform velocity at diffuser slots and suggested that uniform static pressure in diffuser pipes can provide uniform disch
21、arge velocity at slots. The comprehensive reviews and studies on radial and octagonal diffusers were performed by Dorgan and Elleson (Dorgan 1994) and Bahnfleth et al.(2003) and readers are referred to mentioned references for additional details. CFD MODELING The main challenge of the tank sizing wa
22、s to optimize the thermocline in the tanks in order to maximize the output time of stored chilled water while minimizing the height and quantity of the tanks. This was achieved through an innovative procedure in which the tank diffusers were designed through a series of simulations. Reynolds-average
23、d Navier-Stokes equations along with realizable k- two-equation turbulence available in a commercial iterative and control volume base solver were used for all numerical simulation performed in this paper. Figure 2 presents the domain and boundary conditions used for the simulations. Figure 2 Bounda
24、ry conditions set up for thermal storage system. The return flow from the data center enters the first tank at 76 F (24.4 C) with flow rate of 1540 GPM (97.16 L/s). The system provides chilled water at 60 F (15.6 C) at its outlet of last tank. The water inlet temperature was at 76 F (24.4 C) with ve
25、locity inlet based on design flow rate of 1540 GPM (97.16 L/s). The density and viscosity were defined as functions of temperature to be able to capture buoyancy effect. The outlet pressure was considered for boundary condition at the outlet. Intensity and hydraulic diameter were used to specify the
26、 turbulence boundary conditions at inlet and outlet. The recharge flow, roughly 10% of the design flow rate, keeps the temperature of the tank at 60 F (15.6 C) during the non-operational mode. The steady state flow field was used as an initial condition for the transient simulation at time t= 0. The
27、 second order schemes were employed to solve the governing equations and residual accuracy of 1.0e-4 was set up for all variables. The time step was ranged from 0.001 to 1 second to satisfy the residual criteria at each time step and total size of the tetrahedral and hexagonal mesh elements was arou
28、nd 4.4 million cells for each tank. An iterative design methodology was utilized to reduce the flow disturbances in the tank and a three dimensional model used to perform a transient simulation for the system. The strategy for diffuser design included varying the quantity of diffuser rings, the ring
29、 diameter, slot-opening size, and the quantity and variations of slots. Homan and Soo (1997) reported the internal wave motions cause reflective flow and eventually affects the thermocline. A series of baffle rings were mounted in the upper side of the tank to dissipate and eliminate internal wave m
30、otions. With reduced flow disturbance and a more stable thermocline, the tank ton-hour capacity improves and outlet water temperature becomes more reliable. More details on the tank and diffusers optimization strategies can be found in previous authors paper Ghias et al. (2013). BUILD Upon completio
31、n of the tank design after the iterative CFD modeling and simulation process, the next step was to find a reliable manufacturer to fabricate the tanks and the intricate supply and return distribution rings. This step was especially critical given that the slightest variations in ring pipe diameter,
32、slot locations, and slot sizing could have drastic effects on the performance of the tank with respect to the thermocline. Failure to ensure that the tanks and diffusers were fabricated to the exact design tolerances could weaken the credibility of the CFD modeling results. Figure 3 shows a selected
33、 portion of the shop drawing of the custom thermal storage tank. Figure 3 Selected portion of the custom thermal storage tank shop drawing. The thermal storage tanks were custom built by a regional tank manufacturer, with all dimensions provided by the design engineering team. With six (6) thermal s
34、torage tanks in fabrication, the design team visited the production factory at the midpoint of completion of the first tank. A couple major deficiencies in tank construction were found during this visit, one of which is shown in Figure 4. Figure 4 Example of fabrication deficiency found during facto
35、ry inspection. The baffle at the top of the tank designed to limit generating turbulence and eddies in this area, was incorrectly size at half the diameter as designed. Fortunately, the inspection occurred early enough in the production schedule for the fabrication team to correct the first and all
36、the subsequent tanks without disrupting the delivery or installation schedule.Once installed at the site, with final piping connections to the tanks and system filled, the ability to inspect or correct the interior construction would be extremely limited. If the designed tanks were not constructed t
37、o the provided dimensions and specifications, tank performance could be drastically altered from the desired state. This would have affected the final product on two fronts, first with system performance relative to the desired parameters of the project, and second with the validity and accuracy of
38、the actual results versus the predictions of the CFD model. Figure 5 shows the installed thermal storage tanks with final supply and return piping connections. This highlights a key part of the design process often overlooked, and that is verification of the design intent with final fabricated and c
39、onstructed product. Without this process, any deficiencies in performance could be inaccurately attributed to the CFD model and predictions, and could adversely affect whether or not CFD is used as a design tool on future projects. Figure 5 Picture of installed thermal storage tanks with final suppl
40、y and return piping connections. VERIFICATION The commissioning phase of the project, which involved testing the system rigorously to the same conditions previously simulated during design, was very important for a number of reasons. First, the obvious need to validate that the installed system woul
41、d meet the requirements from the owner for fifteen (15) minutes of chilled water at normal supply distribution temperature in the event of a power loss and subsequent loss of active chillers. Second, the design team used the opportunity to verify that the system would meet the design requirements wi
42、thout field modifications or a fine tuning process. Finally, the commissioning process provided the opportunity to use measured data to verify the accuracy of the CFD model predictions. This established an increasing credibility within the design group for use of the CFD as a design tool. In order t
43、o collect the key measurable data to verify the CFD model predictions, specific control points were added to the Building Management System (BMS) that would not otherwise be in the design. For each thermal storage tank, six (6) temperature sensors were installed in 2 ft ( 0.508 m)increments in the v
44、ertical direction of the tank (Figure 6). Collection of this data was used to create a temperature profile throughout each tank, showing the thermocline in the tank as the warmer return water filled each tank in series. Figure 6 Temperature sensors were installed in 2 feet vertical increments to mea
45、sure and verify thermocline of discharging tanks. RESULTS 36:5. Fluent Manual R14. ANSYS Inc. Ghaddar, N. K., Al-Marafie, A. M., and Al-Kandari, A. 1989. Numerical simulation of stratification behavior in thermal storage tanks. Applied Energy, 32 (3): 225-239. Ghias, R., Xu, K., Ellison R., and Eise
46、nhower, C., 2013. Use of Numerical Simulation and Optimization to Analyze the Design and Performance of a Chilled Water Storage Tank, ASHRAE Paper DE-13-C046, 2013. Homan, K., and Soo S. L. 1997. Model of the transient stratified flow into a chilled water storage tank. International Journal of Heat
47、and Mass Transfer, 40(18): 4367-4377. Hudson, H. E., Uhleer, R. B., and Bailey, R.W. 1979. Dividing flow manifolds with square edged laterals. Journal of Environmental Engineering, 105 (4): 745-755. Karim, M. A. 2009. Performance evaluation of a stratified chilled-water thermal storage system. World
48、 Academy of Science, 53: 326-334. Stewart, W., Becher, B. and Cai, L. 1992. Downward impinging flows for stratified chilled water storage. Topics in Heat Transfer, 206 (2): 131-138. Yoo, J., Wildin, M. W., and Truman, C.R. 1986. Initial formation of a thermocline in simplified storage tanks. ASHRAE
49、Transaction, 92 (2): 280-292. Zhai, Z. 2006. Application of Computational Fluid Dynamics in Building Design: Aspects and Trends. Indoor and Build Environment, 15:4:305-313. Zurigat, Y., Liche, P., and Ghajar, A. 1991. Influence of inlet geometry on mixing in thermocline thermal energy storage. International Journal of Heat and Mass Transfer, 34: 115-125.
