Chilled water plant retrofit--a case study.

ABSTRACT

This paper outlines a phased retrofit plan for the central chilled water plant at Sonoma State University. The plant has experienced many operational problems and is composed of equipment (chillers, pumps) that has operated well beyond its expected service life. Furthermore, the campus master plan calls for the addition of several future buildings that will tax the already marginal capacity of the system. The retrofit plan, which includes converting the distribution system to primary-secondary/series-tertiary pumping, will enhance the plant capacity to serve future buildings, resolve existing hydronic distribution system problems, improve energy efficiency, reduce maintenance costs, and improve system reliability.

INTRODUCTION

The central chilled water plant at the Sonoma State University campus in Rohnert Park, California, has experienced many problems that are common to large plants serving multiple buildings, including degrading temperature difference (often called "low [DELTA]T syndrome") and low flow rates at remote buildings. The campus master plan also calls for new construction that will gradually add load to the plant over the next several years. This paper discusses a study and phased retrofit plan designed to resolve operational problems and increase plant capacity with minimal cost investment.

CURRENT PLANT DESIGN

Figure 1 is a schematic of the existing plant. CH-1 and CH-2 date back to 1965. CH-2, which has not been operational for many years, is in the process of being replaced. CH-3 was added in the 1976. In 1996, a stratified thermal energy storage (TES) tank was added to the system. The system was designed for "full storage," meaning the tank is large enough to supply the campus load for the entire utility on-peak window (noon to 6 p.m.). The tank is charged during off-peak periods by opening valve V-1 and running one of the primary pumps and chillers (typically CHWP-4 and CH-3, which are the newest and most efficient). Valve V-2 is modulated to bypass water back to the chiller to reduce the temperature difference ([DELTA]T) across it. This was required since the primary pumps are constant speed and the chillers were originally selected for a 12[degrees]F [DELTA]T, while the system (and tank) [DELTA]T increased from 12[degrees]F to 15[degrees]F or 16[degrees]F after the TES was installed and design supply water temperature was reduced from 43[degrees]F to 40[degrees]F. (1) At a 16[degrees]F [DELTA]T, the TES has a capacity of roughly 7200 ton-hours.

TES water is distributed to buildings with a single distribution pump (CHWP-5 in Figure 1). The pump has a variable-frequency drive (VFD) that is controlled to maintain differential pressure measured about halfway down the distribution mains.

Buildings are connected to the central plant using a variety of "bridge" connections. As is typical of many campuses, there is no clear campus standard for the design of central plant connections, so each building is different. Figure 1 shows the three basic connection types, although there are many variations. The connections shown for typical buildings A and C are similar: they are standard primary-secondary bridges except that building A has a manual valve (normally open) in the common leg. The connection shown for building B, which was used at three adjacent campus buildings, uses a three-way valve rather than a standard primary-secondary connection.

[FIGURE 1 OMITTED]

The buildings also are mixed with respect to distribution within the buildings. Most (unfortunately) are constant volume with three-way valves at all coils. A few use two-way valves with constant-speed, variable-flow pumps riding their pump curve. Only one building, the most recently built, uses two-way valves on most coils and includes variable-frequency drives on the building pumps.

Control systems are also mixed among two direct digital control (DDC) system vendors in the buildings and an industrial programmable logic controller (PLC) system in the central plant. In some buildings, bridge controls are performed by stand-alone electronic controllers and by pneumatic controls. None of the control systems "talk" to one another, so building cooling and heating demands are not known to the central plant. Accordingly, the central plant is largely manually controlled, with all equipment started and stopped by operators. The TES charge mode is also manually controlled.

Figure 2 shows the central plant and chilled water distribution system to campus buildings. The 14-inch chilled water main is run in a tunnel under one of the main walking arteries between buildings. Branches to buildings are generally direct-buried piping, originally Transite but gradually replaced with insulated steel to repair leaks.

The current 15-year master plan calls for the addition of almost 400,000 [ft.sup.2] of space, including instructional buildings, a student union building, and a recreation center. The estimated added cooling load is approximately 810 tons and 925 gpm. The relatively low added flow rate is due to a change in coil design conditions to 21[degrees]F or higher [DELTA]T to reduce chilled water pump energy, pipe, and pump sizes and to enhance the capacity of the TES tank.

Over the last few years, two buildings have been taken off the plant. They are cooled instead by indirect-direct evaporative cooling (IDEC) systems. The systems have proven to be effective for these two projects and are currently being considered for some of the future buildings instead of connecting them to the central plant. The life-cycle costs of using IDEC vs. expanding plant capacity will be compared as part of an ongoing chilled water plant master plan.

Plant loads (including projected loads) are summarized in Table 1, while plant capacity is summarized in Table 2. Building design loads were extracted from design drawings where they existed (not surprisingly, many were not) or pump nameplates. Where these data were not available, loads were estimated based on building size and type. A diversity factor of 65% was used based on experience and to match operator's observations. Unfortunately, no records are kept of actual plant load, and the PLC system's trending capability was not operational.

From the two tables, it appears that the plant's capacity is already marginal (1253 tons estimated load vs. 1200 tons capacity and 1849 gpm vs. 1800 gpm capacity). Capacity clearly must be increased to handle future loads added to the plant over the next 15 years.

CURRENT SYSTEM OPERATING PROBLEMS

Problems and issues with the current central plant include the following.

1. Poor reliability due to aging central plant equipment and lack of redundancy. One chiller and one tower have not been operational for years; the others are 30 to 40 years old. Pumps and other plant equipment are similarly aged and in poor operating condition. The plant currently has a single distribution pump. With little or no redundancy and extremely old equipment, there is a significant exposure to plant failure.

[FIGURE 2 OMITTED]

2. Inability to charge the tank and serve loads simultaneously. The primary pumps (originally 100 feet head) were trimmed to serve the relatively small tank charging circuit head and thus cannot run simultaneously with the distribution pump (CHWP-5 in Figure 1). This poses significant disadvantages since it means that the plant cannot serve 24/7 loads, such as data centers and other IT rooms. Instead, these spaces have dedicated air-cooled AC units or chillers, both of which increase installed costs, building space requirements, maintenance costs, and energy costs.

3. Three-way valves. There is almost nothing that can sabotage a central plant, in particular one with a TES system, more effectively than the use of three-way valves. Almost all of the buildings served by the plant have them, and most have only three-way valves. Three-way valves at coils that are partially loaded (the predominant condition) constantly feed cold, chilled water into return lines. This increases pump energy, reduces effective plant capacity by reducing its ability to take advantage of load diversity, and reduces the capacity of the TES tank.

4. Improper bridge controls. The control valves in the building bridge connections are being controlled to maintain the return water temperature at a minimum of 55[degrees]F. If return water is too cold, it is essentially recirculated back into the building to absorb more heat. While intuitively it may make sense that recirculating water will increase return water temperature, this is not usually the case. Table 3 shows that for a given load, increasing entering chilled water temperature results in a lower, not a higher, leaving return water temperature. This is not as counterintuitive as it may seem at first: it makes sense that the coil will need less water as the supply temperature drops, and it makes sense that as the water flow rate is reduced, the water has more time in the coil to absorb heat. So recirculating water not only increases the flow in the building loop, it also may increase flow in the secondary loop. Furthermore, this control logic, particularly in buildings with three-way valves, almost surely will result in starving building loads. Clearly, with the coil in Table 3, if the system started with the return water below the 55[degrees]F setpoint, the controls would cause water to recirculate, which makes the return water colder, which in turn makes the bridge valve close further. This "death spiral" ultimately results in the valve being fully closed with no loads fully satisfied. Without any plant chilled water being supplied to the building loop, return water temperature will begin to slowly rise since coil entering air temperature is above 55[degrees]F. Once the return water rises above 55[degrees]F, only enough chilled water will be injected into the loop to keep it at that temperature. Because the 55[degrees]F setpoint is relatively low, in practice this control logic in buildings with predominantly two-way valves has not been a problem: the return water temperature is pretty much always above 55[degrees]F, so the control logic simply does no thing. But with buildings that have three-way valves, the "death spiral" results in the bridge chilled water valve fully closing for extended periods. Because of the undercooling problems this caused, logic in these buildings was revised to limit how warm the chilled water supply temperature to the building was allowed to rise regardless of return temperature; e.g., the supply water temperature was limited to 45[degrees]F. While this may mitigate the failure of the bridge control logic, Table 3 demonstrates that it still results in higher chilled water flow in the building and primary circuits and cooler, not warmer, return temperatures.

5. Lack of control system intercommunication and poorly operating controls. Building control systems are largely stand-alone. There is no ability to send demand information from the buildings to the plant, e.g., to allow pump differential pressure setpoint reset. In many cases, building pumps and chilled water valves never shut off. This requires that the plant be manually valved off from the distribution system at night so that the tank can be charged without simultaneously cooling unoccupied buildings. In fact, the entire central plant must be manually controlled due to the lack of feedback from building systems and due to inoperative central plant controls and automatic equipment isolation valves.

6. Series bridge connections with constant-speed building pumps. Three buildings have connections like that shown for building B in Figure 1. This design causes the building pump to be in series with the plant distribution pump. Since the pumps are constant-speed and the buildings use three-way valves, these buildings use more water than design, potentially starving other buildings. The speed of the plant distribution pump is controlled to maintain a very low differential pressure (0.5 psi) at a position upstream of these buildings. At times, this has caused differential pressure and flow to reverse in chilled water mains downstream of the buildings.

7. Poor energy efficiency. Chillers have design efficiencies of almost 0.7 kW/ton; surely at their advanced age actual efficiency is even worse. Primary pumps, originally selected for 100 ft of head, were retrimmed to operate at about 35 ft when the TES was installed, which results in their operating at an inefficient point on their curves. The original installation included orifice plate flow meters, which have a high pressure drop and are no longer operational.

OPTIMIZED PLANT DESIGN

A good approach to retrofit work is to start with what the ideal design might look like and then see how the system can be reconfigured to best approach that ideal. Figures 3 and 4 show what might be the most life-cycle-cost effective designs for a campus central plant with TES. Key features of the designs are the following.

* No three-way valves. There are few, if any, good reasons to use constant-flow (predominantly three-way valve) systems in these buildings; variable-flow (predominantly two-way valve) systems are less expensive to install, use less energy (even without variable-frequency drives on pumps), and are arguably self-balancing (Taylor 2002a, 2002b). Furthermore, with perhaps one exception, three-way valves should not be installed anywhere in a variable-flow system. Three-way valves are sometimes installed at the end of each branch (e.g., building A in Figure 1) to keep water constantly moving through the circuit, thereby ensuring that chilled water will be available immediately on demand by any coil in the system. The concept probably comes from domestic water recirculating systems, which are designed to make sure people do not have to wait to get hot water at a lavatory. But the concept is not applicable to most chilled water systems simply because it seldom matters if coils have to wait a few seconds or a few minutes to get chilled water when they call for it. For instance, assume a system has 250 ft of piping from the building pump to the farthest valve. At typical piping velocities, it would take less than a minute for water to travel from the pump to the coil. Surely the space will not overheat in such a short time period. Hence, there truly is no need for end-of-the-line three-way valves. The one application where three-way valves might logically be used is when coils are not controlled by the DDC system (so their position is not known). In this case, one or perhaps two three-way valves should be installed to ensure pumps are not deadheaded. Running a pump deadheaded for a long period of time can cause seal failures as the temperature of the water increases. But only a small flow through the pump is required to prevent this problem (Rishel 1996), so no more than one or two three-way valves are required per system. In Figure 3, no three-way valves are used at all because in an ideal application, valve position will be known through the DDC system, so pumps will simply be shut off when all valves are closed.

[FIGURE 3 OMITTED]

* High [DELTA]T coils. Building coils would be selected for very high [DELTA]T to reduce piping and pump costs and pump energy and to enhance the capacity of the TES. Lifecycle cost studies (Taylor et al. 2000) have shown that life-cycle costs tend to be optimized at the highest practical [DELTA]T because the increase in coil costs and air-side pressure drop tend to increase very slowly relative to the benefits of high [DELTA]T on the chilled water system. Return water temperature is limited by the coil entering air temperature and coil size limitations. The new University of California campus at Merced has a coil design leaving water temperature criteria of no more than 10[degrees]F below the coil entering air temperature, or 75[degrees]F, whichever is lower. So typical coils with a 75[degrees]F entering air temperature would be sized for a 65[degrees]F leaving water temperature. At 40[degrees]F entering water temperature from the TES, the [DELTA]T at UC Merced is expected to be at least 25[degrees]F.

* No bridge connection. Table 3 suggests that the best plant performance will result when the supply water temperature to buildings is as cold as the plant can deliver: flow rate will be the lowest and return temperature will be the highest. This, in turn, suggests that there should be no primary-secondary bridge connection (no common leg) at the buildings; simply pipe the building pumps directly to supply and return mains.

* Common leg at plant. Locating the common leg at the central plant rather than at building bridge connections eliminates the need for plant distribution pumps (effectively, CHWP-5 in Figure 1 is eliminated). The building pumps themselves provide all of the head required to move water from the plant to the buildings and back. A distribution pump at the plant (as in Figure 1 and also shown in Figure 5) would have to provide sufficient head to serve the worst-case building, which results in excess head at hydraulically closer buildings. This extra head is wasted by the building bridge connection control valves. The differences in pump brake horsepower (BHP) and motor horsepower (MHP) are shown in Table 4 for a system with a distribution pump at the plant and common legs at each building bridge connection (as in Figure 1) compared to a system with the common leg at the plant (Figure 3) using actual design data from the Sonoma State plant. Pump head in each case was calculated using a commercially available piping simulation program. The only difference between the two simulations is the location of the common leg; the Figure 3 design will also have lower head by eliminating the pressure drop from the bridge connection (control valve, additional isolation valves, and tees) and primary pump piping (isolation valves, check valve, strainer, etc.), which is conservatively ignored. BHP and MHP were determined using realistic pump selections from a manufacturer's pump selection program. In addition to lower pump energy, the system in Figure 3 will have lower first costs due to the elimination of the secondary pumps and multiple bridge connections and associated controls. One disadvantage of this design is that pumps at smaller buildings can be low flow/high head pumps requiring either 3500 rpm motors or pumps in series. The higher-speed pumps are less expensive but may cause noise problems depending on the location of piping and sound/vibration isolation measures.

* TES in common leg. Placing the TES in the common leg allows it to be charged, discharged, and operated simultaneously with chillers without operating any valves. This is an extremely simple, flexible, and inexpensive approach.

* High-efficiency chillers. Chillers with a TES system should be optimized for charging duty, which generally occurs at night when wet-bulb temperatures (and condenser water temperatures) are low but must be capable of operating during daytime conditions as well in case of TES failure or if simultaneous on-peak (partial storage) operation is required. Three chillers are shown in Figure 3, each sized for 50% of the tank charging load (n + 1 redundancy). The number of chillers that will provide redundancy while optimizing first costs and energy efficiency will vary with plant size. A more expensive but more efficient chiller plant variation is shown in Figure 4. Chiller efficiency is primarily a function of leaving water temperature, not entering water temperature, so piping two chillers in series reduces overall kW/ton compared to piping chillers in parallel. The three chillers shown in Figure 4 are piped so that any two chillers can operate in series. The chillers can also operate alone, which is valuable for partial storage operation.

* Chiller capacity control for partial storage operation. If the TES is not large enough to handle the campus peak load or integrated load over the utility on-peak period (12 noon to 6 p.m. in this case), the chiller must be capable of operating along with the TES. This is called a partial storage design. There will be days, other than perhaps the design cooling day, when the chiller must run to augment the TES but not at full capacity. To allow part-load chiller operation, bypass valve V-1 in Figure 3 must be installed. Without it, if a chiller and primary pump are operated along with the TES, the chiller will be fully loaded with the TES providing the trim, not the other way around. This is because the chiller will always see the same flow (constant-speed primary pump), entering water temperature (that returning from the buildings), and leaving water temperature (39[degrees]F to 40[degrees]F as required to maintain full TES charge). To allow the TES to be fully loaded while the chiller provides the trim, we must be able to unload the chiller. This is done by bypassing supply water through V-1, reducing chiller entering water temperature. The location of the bypass is critical; it cannot be located on the opposite side of the TES or water returning to the TES will also be cooled. Energy performance can be improved by adding variable-frequency drives to the primary pumps, as shown in Figure 4. This allows primary flow rate to be reduced, reducing chiller capacity and pump energy. Once the minimum chiller flow rate is reached, further chiller unloading is accomplished as before, by modulating bypass valve V-1.

[FIGURE 4 OMITTED]

PLANT REVISIONS

Plant upgrades are focused on the following key issues.

* Enhancing the plant capacity for future buildings projected in the 15-year master plan

* Resolving existing hydronic distribution system problems

* Reducing system energy costs

* Reducing maintenance costs

* Improving system reliability

Enhancements were, as always, constrained by available funding, which in this case is spread over more than one fiscal year. Hence, recommended retrofits had to be cost conscious, life-cycle cost effective, and capable of being phased in over a two-year period due to limited funding.

While many plant retrofit measures were recommended and implemented (including replacement of chiller, pumps, controls, etc.), this paper focuses only on the revised hydronic distribution system.

Ideally, the existing plant in Figure 1 could be converted inexpensively to the plant shown in Figure 3. But immediately we come across a complication: The building pumps in Figure 3 must have sufficient head to pump water from the central plant through the building and back. In this case, pump heads must be 45 to 75 ft higher than the existing pumps are sized for. The cost to replace all of the building pumps, pump motors, variable-frequency drives, electrical service, etc., would have been prohibitively expensive. Without this change, plant distribution pumps were still required.

However, the need for distribution pumps does not mean that bridge connections and common legs are required. Instead, a scheme like that shown in Figure 5 can be used. The building (tertiary) pumps are in series with the plant distribution (secondary) pumps. It was noted previously that series pumping like this caused problems with excess flow through some buildings while starving others. But that was because the building pumps were constant speed. In our ideal design, building pumps will have VFDs. If the secondary pump pressure causes the building pumps to overpressurize, the building pumps will simply slow down. Some engineers have expressed concern that overpressurization may still occur due to the need for a minimum VFD speed (e.g., 20 Hz) for motor cooling, but VFD and motor manufacturers have confirmed that high minimum speeds are not required for motor cooling in cube-law (power proportional to the cube of pump speed) VFD applications such as pumps and fans. On the authors' projects, minimum speed is determined in the field as that required to cause the pump to overcome friction and begin to rotate, typically only 6 Hz or so. We have not experienced a single motor failure with this approach.

But even if overpressurizing (forcing control valves open) is not an issue, pressure may be more than needed even at minimum building pump speed for buildings close to the plant. In this case, the highest differential pressure (DP) created by the secondary pumps at any building is calculated to be about 30 ft, not enough to supply design flow but more then enough for part-load operation. So that building coil control valves do not have to throttle excess pressure and waste pump energy, a low pressure drop, check-valve bypass around the pumps can be provided for buildings close to the plant, as shown in building A in Figure 5. This will allow the building to operate at low loads without having to run the building pumps, saving energy and also mitigating the risk of building pump failure for those buildings with single pumps. Finally, buildings can include a control valve on the supply for further differential pressure control if needed. (This valve is also used for building shutoff when the building is not in service. It is technically not required if all building coil valves are two-way valves controlled to shut off when not in service, but it is inexpensive insurance to ensure that buildings never waste chilled water and never are overpressurized.) With the piping scheme shown for building A, the building pressure control sequences the valve and building pump speed based on a control loop measuring differential pressure at the end of the building piping loop, as shown in Figure 6. When cooling is required, the valve opens as required to maintain building DP at setpoint using the available pressure from the secondary pumps. At 50% control loop output, the valve is fully open and the building pump starts. Pump speed ramps from minimum speed (e.g., 6 Hz) up to full speed as load in the building increases. As the load decreases, the pump will reduce to minimum speed, eventually shutting off again (with a small deadband to prevent pump short-cycling, represented by dashed lines in Figure 6).

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

So now the compromise goal is to convert the plant in Figure 1 to that in Figure 5 in a phased basis that simultaneously addresses other plant needs such as replacing old equipment. The following phasing scheme was proposed and is currently being implemented: (2)

1. Phase 1: Replace CH-2, CHWP-1, and CHWP-2. The work in this phase is highlighted in red in Figure 7 and includes:

* Demolish inoperative chiller CH-2, two of the existing 1960s-era chilled-water pumps, and one cooling tower cell (not shown in figures).

* Install a new 750 ton chiller selected based on a performance-based bid for minimum life-cycle cost. Because of the TES, the lowest life-cycle-cost chiller is a fixed-speed chiller; chillers with variable-speed drives did not prove to be cost-effective since the chiller runs at nearly full load whenever it is on. The 750 ton capacity was selected because it fit within existing constraints (size, electrical service, pipe sizes) and is large enough to charge the tank during off-peak hours while simultaneously serving a 200 ton 24/7 load that is expected to be added to the system in the future once the plant is capable of serving loads and charging the TES simultaneously. The exact details of the sizing analysis are beyond the scope of this paper.

* Install two new 1000 gpm primary pumps, one for standby operation, selected for the charging circuit head (~35 ft). The charging circuit is from the pumps through the chillers, through valve V-1 to the storage tank, then back to the pumps. The original pumps were sized at 100 ft for distribution duty, but the pumps are no longer used in that manner even with the current plant design.

* Install a new 1800 gpm, 100 ft head distribution pump in parallel with existing pump for redundancy and future capacity enhancement. Calculations indicate that the existing pump head of 100 ft is adequate to serve all buildings included in the 15-year master plan.

[FIGURE 7 OMITTED]

2. Phase 2: Convert building systems to variable flow, variable speed. The work in this phase is highlighted in Figure 8 and includes:

* Close off the bypass lines where three-way valves are used. In some cases, this may require replacing the valve and/or actuator if the existing valve does not have sufficient shutoff head. Valves will be tested for shutoff capability and controllability once the bypasses are closed.

* Add VFDs to pumps and associated DP controls. In some cases, motors may be replaced as well if the manufacturer indicates the motor is not compatible with VFDs.

* Install pump bypass with check valve on building pumps located near the plant (three buildings).

* Shut off decoupler lines at bridge connections by shutting existing valves (for connections like building A in Figure 8), by fixing three-way valves into the straight-through position (building B in Figure 8), or by installing a normally closed valve (building C in Figure 8). (The decoupler is not removed entirely because it is useful for allowing the building pump to recirculate water within the building for cleaning after the system is modified for tenant changes in the future.)

[FIGURE 8 OMITTED]

* Modify the controls for the new building valve and pump speed control logic (Figure 6) and upgrade controls to allow communication to the central plant control system for optimized secondary pump operation.

3. Phase 3: Convert plant to allow simultaneous TES and chiller operation. The key to this phase is effectively moving the TES discharge pumps (CHWP-5 and the new distribution pump) from the common leg to the secondary loop to create a true primary-secondary arrangement with only the TES in the common leg. The work in this phase is highlighted in Figure 9 and includes:

* Repipe the system as indicated in Figure 9. This is needed in order for the TES tank to become the common leg between the primary and secondary (distribution) pumps.

* Valve off the existing return line to the chillers as shown in Figure 9 and permanently open valve V-1 (convert it from an automatic to a manual valve).

4. Phase 4: Replace CH-1 and CHWP-3. Because of their already advanced age, CH-1 and CH-3 are not very reliable backups to new chiller CH-2. This phase includes replacing the older of the two chillers, CH-1, with a new 750 ton chiller similar to CH-2 and replacing a primary pump with a new 1000 gpm pump similar to new pumps CHWP-1 and CHWP-2.

5. Phase 5: Replace CH-3 and CHWP-4. The new chiller and pump should be sized for 1350 tons capacity. The TES tank discharge rate is 1350 tons and the distribution system has 2700 tons capacity, so replacing CH-3/CT-2 with equipment sized for 1350 tons will "max out" the current distribution pumping system and provide full n + 1 redundancy (TES may be down with plant at full capacity). It also allows more flexibility for additional 24/7 cooling systems to be served by the central plant while simultaneously charging the TES tank. Because the new 1350 ton chiller will be more efficient than the existing 1000 ton chiller, plant infrastructure revisions should not be required.

Once Phase 5 is completed, the plant will have the capacity shown in Table 5. Capacity is based on the assumption that all new buildings will have coils selected for at least 20[degrees]F [DELTA]T, increasing average plant [DELTA]T from 15[degrees]F to 18[degrees]F. The total distribution system capacity will be 3600 gpm and 2700 tons, more than the 2772 gpm, 2061 tons required by all proposed buildings through 2020 (Table 1). This will be met on the primary side by a combination of the TES tank and two of the three chillers using partial-storage logic (simultaneous TES discharge and chiller operation) during the utility on-peak period. Partial-storage logic is much more complex than full-storage logic, so at some point over the next few years, an additional study will be performed to see if increasing TES capacity is feasible (there are significant space constraints) and cost-effective.

SUMMARY AND DISCUSSION

The central chilled water plant at Sonoma State University has experienced many operational problems and is composed of equipment (chillers, pumps) that has operated well beyond its expected service life. Furthermore, the campus master plan calls for the addition of several future buildings that will tax the already marginal capacity of the system. A phased retrofit plan was developed, including converting the distribution system to primary-secondary/series-tertiary pumping, that will enhance the plant capacity to serve future buildings, resolve existing hydronic distribution system problems, improve energy efficiency, reduce maintenance costs, and improve system reliability.

[FIGURE 9 OMITTED]

REFERENCES

Rishel, J.B. 1996. HVAC Pump Handbook. New York: McGraw-Hill.

Taylor, S.T. 2002a. Balancing variable flow hydronic systems. ASHRAE Journal 44(10):17-24.

Taylor, S.T. 2002b. Degrading chilled water plant delta-t: Causes and mitigation. ASHRAE Transactions 108(1):641-653.

Taylor, S., P. DuPont, M. Hydeman, and C. Eley. 2000. Cool-Tools[TM] Energy Efficient Chilled Water Plant Design and Performance Guide. San Francisco, CA: Pacific Gas & Electric Company.

Steven T. Taylor, PE

Fellow ASHRAE

Steven T. Taylor is a principal at Taylor Engineering, LLC, Alameda, California.

(1.) Typical cooling coils that were sized for a 10[degrees]F or 12[degrees]F [DELTA]T at 43[degrees]F to 45[degrees]F entering water temperature can achieve on the order of 15[degrees]F to 16[degrees]F [DELTA]T at 40[degrees]F entering water temperature. See Table 3 for an example.

(2.) Cooling towers and condenser water pumps are also proposed to be replaced, but this paper focuses on the chilled water system only.

Table 1. Plant Loads

                        Design Flow  Design Load
Building                (gpm)        (tons)

Stevenson                480          300
Ives                     295          185
Physical Education        45           28
Student Health Center    109           68
Nichols/Carson           312          195
Zinfandel Res/Dining      64           40
Commons Cafeteria        115           72
Student Union             96           60
Fine Arts                290          218
Theater                  130           81
Schulz Info Center       908          681
Current sum-of-peaks    2844         1928
Diversity factor          65%          65%
Current plant load      1849         1253
Projected future loads   923          808
Total future loads      2772         2061

Table 2. Current Plant Capacity

                                      Design Flow  Design Load
Component                             (gpm)        (tons)

TES Tank
Capacity (at 16[degrees]F [DELTA]T,   685,000 gal  7200 ton x h
  95% full)
Discharge rate (CHWP-5)                  1800      1200
Charge rate                              1500      1000

Chillers
CH-1                                     1200       610
CH-2                                  (not operational)
CH-3                                     2000      1000

Table 3. Coil Performance with Increasing Chilled Water Supply
Temperature*

Entering                                         Leaving
Chilled Water            Flow Rate,  Delta-T,    Chilled Water
Temperature, [degrees]F  gpm         [degrees]F  Temperature, [degrees]F

40                        39         16.3        56.3
42                        47         13.4        55.4
44                        62         10.1        54.1
46                        96          6.6        52.6
47                       136          4.6        51.6

* Based on a 96 x 30 eight-row 80 fpf coil, 78[degrees]F entering
dry-bulb/62[degrees]F entering wet-bulb, 53[degrees]F leaving air
temperatures.

Table 4. Design Pump Power as a Function of Common Leg Location

          Common Leg at         Common Leg at
Pumps at  Building Bridge       Central Plant
Building  Total BHP  Total MHP  Total BHP  Total MHP

A           3.30       4          9.68      15
B           7.24      10         13.98      20
C           1.50       2          4.46       6
D           1.14       2          1.38       2
E           2.50       4          5.7       10
F           4.16       6          9.68      15
G          35.5       40         53.6       80
H           6.08      10         16.6       20
I           1.48       2          4.60       6
J           2.44       4          5.40      10
K           0.88       1.5        3.20       6
Plant      90.40     120
          156.6      205.5      128.3      190
          Percent reduction:     18%         8%

Table 5. Final Plant Capacity

                                     Design Flow  Design Load
Building                             (gpm)        (tons)

TES Tank
Capacity (at 18[degrees]F [DELTA]T,  685,000 gal  8100 ton x h
  95% full)
Discharge rate                          1800      1350
Charge rate                             1800      1350
Distribution pumps                      3600      2700

Chillers
CH-1                                    1000       750
CH-2                                    1000       750
CH-3                                    1800      1350