Printer Friendly

Performance of a demand-limiting control algorithm for hybrid cooling plants.

INTRODUCTION

A hybrid chiller plant employs a combination of chillers that are "powered" by electricity and natural gas. Chillers that employ natural gas include absorption and engine-driven chillers. Wylie and Alvarez (1997), Meckler (1997), Arnold and Banfleth (1998), Nowakowski and Gramlich (1999), and Smith (2002) presented overviews of hybrid chiller plants, including their advantages and disadvantages. The major advantage of using natural gas chillers in hybrid plants is a reduction in peak demand and on-peak energy usage associated with electricity, which can reduce overall operating costs. The electric demand cost can often account for about half of the total air-conditioning bill. In order for hybrid plants to be economically viable, on-peak electrical demand and energy cost savings must be sufficient to offset higher initial and maintenance costs. In addition, it is important to have effective supervisory control strategies that identify the best chiller sequencing and loading and cooling tower fan control in response to changes in utility rate periods and operating conditions.

Although there is a large body of literature related to supervisory control for all-electric plants, there is very little literature on supervisory control of absorption, engine-driven, and hybrid chiller plants. For all-electric plants, chapter 41 of the 2003 ASHRAE Handbook--HVAC Applications includes a number of intelligent control strategies, including strategies for sequencing and speed-control for cooling tower fans, resetting of chilled-water supply setpoints for systems with fixedspeed or variable-speed pumping, and sequencing and loading of chillers. These strategies are all based on minimizing plant power consumption at any instant in time.

The problem of determining optimal control settings for hybrid cooling plants is considerably more complex and should include the effects of electrical and gas energy, peak electrical demand, and maintenance costs. Braun (2006, 2007a, 2007b) has developed control strategies that include these effects. The current paper addresses implementation of the strategies and evaluates cost savings potential associated with the demand-limiting control algorithm. The cost savings are evaluated for a range of hybrid systems using a simulation tool described by Braun (2006).

SYSTEMS

Figure 1 gives a general schematic of the chiller plant configuration that is considered. Multiple chillers, pumps, and cooling tower cells are employed and controlled with a goal of meeting the required plant cooling requirements in the most cost effective manner.

[FIGURE 1 OMITTED]

Table 1 gives common parameters that were employed for the simulations presented in this paper. Four different chiller types are considered within the simulation tool. Chiller energy performance is characterized in terms of rated COPs and manufacturers' part-load characteristics that are embedded within the tool for the different chiller types. The chiller maintenance costs are presented in terms of costs per unit of runtime and rated capacity and were derived from average annual maintenance costs for electric, direct-fired absorption, and engine-driven chillers given by GTI (2005). The cooling capacity of the chiller plant is determined by specification of the individual chillers. The cooling tower size and design fan airflow rates and power consumption are scaled according to the plant heat rejection requirements. Rated airflows and power are specified in terms of volumetric airflow rate per unit condenser heat rejection rate at design (cfm/ton or L/s.kW) and fan power per unit volumetric flow rate (W/cfm or W.s/L). The design ambient conditions, performance ratings, and embedded component performance characteristics are used along with a specification of the number, type, and cooling capacities of chillers to determine the cooling tower design airflow and fan power requirements.
Table 1. Common Parameters for Hybrid Cooling Plant Simulation Studies

Parameter Value

Design Conditions
 Wet-Bulb Temperature 80[degrees]F
 Dry-Bulb Temperature 95[degrees]F

Electric Chillers
 Rated COP 6.0
 Condenser Water Flow Rate 3 gpm/ton
 Condenser Water Pump Power 15 W/gpm
 Maintenance Costs 0.008 $/ton-h

Absorption Chiller
 Rated COP 1.0
 Condenser Water Flow Rate 4 gpm/ton
 Condenser Water Pump Power 15 W/gpm
 Maintenance Costs 0.01 $/ton-h

Engine-Driven Chiller
 Rated COP 1.5
 Condenser Water Flow Rate 3 gpm/ton
 Condenser Water Pump Power 15 W/gpm
 Maintenance Costs 0.016 $/ton-h

Cooling Tower
 Motor Variable-Speed
 Rated Airflow 200 cfm/ton
 Rated Fan Power 0.4 W/cfm

Utility
 On-Peak Electric Period 10 a.m.-10 p.m., M-F


Table 2 gives the matrix of systems and utility parameters used for evaluating cost savings associated with the demand-limiting control strategy. The simulations were performed for all combinations of the buildings, locations, chillers, and utility rates specified in Table 2 and using the parameters in Table 1. Three different building types (office, school, and hospital) were considered in three locations (Chicago, Houston, and San Francisco). The office building has a relatively flat cooling load profile during the occupied period but has practically no load during unoccupied periods. The school has a relatively flat load profile during the school day and additional loads that occur during periods of activities on evenings and weekends. The hospital is occupied 24 hours a day and has much less variation in cooling loads over a typical day as compared with the office and school. Hourly plant cooling loads for the different buildings and locations were determined using DOE-2. TMY2 weather data (NREL 1995) were used for the cooling load and hybrid plant simulations. The simulated cooling load data were scaled so that the specified plant cooling capacities were 20% greater than the peak load. A detailed description of the simulated buildings is given by Braun (2006) along with sample load profiles.
Table 2. Simulation Matrix for Evaluation of Hybrid Plant
Control Strategies

Parameter Range of Values

Buildings Office, School, Hospital

Locations Chicago, Houston, San Francisco

Chillers 1. 500-ton Variable-Speed Electric, 500-ton
 Absorption

 2. 500-ton Fixed-Speed Electric, 500-ton Engine

 3. 500-ton Fixed-Speed Electric, 250-ton Absorption

 4. 500-ton Fixed-Speed Electric, 500-ton Absorption,
 500-ton Engine

 5. 500-ton Fixed-Speed Electric, 500-ton
 Variable-Speed Electric, 500-ton Absorption,
 500-ton Engine

Electric Energy 1. 0.10 $/kWh On-Peak, 0.05 $/kWh Off-Peak
Rates

 2. 0.15 $/kWh On-Peak, 0.075 $/kWh Off-Peak

 3. 0.05 $/kWh On-Peak, 0.025 $/kWh Off-Peak

Electric Demand 1. 15 $/kW On-Peak, 0 $/kW Off-Peak
Rates

 2. 25 $/kW On-Peak, 0 $/kW Off-Peak

 3. 5 $/kW On-Peak, 0 $/kW Off-Peak

Gas Rates 1. 0.80 $/therm

 2. 1.20 $/therm

 3. 0.40 $/therm


CONTROL STRATEGY DESCRIPTION AND IMPLEMENTATION

A key element of the strategies is the use of targets for plant power demand. The targets are constraints on peak plant power demand that are established for each type of utility rate period (e.g., summer on-peak periods) to minimize either the billing period energy or demand costs, whichever is more economical. During the course of the billing period, chiller and cooling tower strategies are employed to minimize combined chiller and maintenance costs subject to the demand constraint. Figure 2 gives an overview of the individual elements that compose the overall control strategy. These elements would be executed at regular decision intervals, such as every five minutes. Descriptions of each control strategy are provided in following subsections.

[FIGURE 2 OMITTED]

Updating Chiller Sequencing Lists

The primary method of limiting demand is to change the sequencing of chillers. Sequencing defines the order for bringing chillers online and offline. The process of determining chiller sequencing is handled by defining separate sequencing lists for different stages of demand limiting. In the absence of a demand constraint or if the current plant demand is well below the demand target, then the chillers should be brought online in order of decreasing combined energy and maintenance cost per unit of cooling. This is defined as the 0th stage of demand limiting. If the plant power consumption is close to the demand target, then the demand-limiting (i.e., gas) chiller with the lowest energy/maintenance cost should move to the top of priority list and the other chillers should move down. This is the first stage of demand limiting. Additional stages of demand limiting are enabled as necessary to keep the plant power consumption below the current target. The demand-limiting stages are sequentially disabled in reverse order when the power consumption falls significantly below the demand target.

The sequencing lists for the different demand-limiting stages can change with time due to changing utility rates. In general, the list for the kth stage of demand limiting is created from the k-1th stage list by moving the kth most cost-effective demand-limiting chiller to the kth position and moving the other chillers down. In total, the number of stages of demand limiting is equal to the number of gas chillers, [N.sub.g]. The list for the 0th stage of demand limiting is created by using the chiller operating cost per unit time and design cooling capacity ($/ ton-h) at the rating condition determined with

[j'.sub.ch, rated, i] = [ER.sub.i]*[1/[COP.sub.rated, i]] + [MR.sub.i] + [ER.sub.e].[P'.sub.cwp, rated, i], (1)

where [ER.sub.e] is cost per unit of electrical energy input and for the ith chiller, [ER.sub.i] is cost per unit energy input (heat or electricity), [COP.sub.rated,i] is COP at the rating condition, [MR.sub.i] is a chiller maintenance cost per unit of runtime and rated capacity, and [P.sub.cwp,rated,i] is condenser water pump power per unit of rated chiller cooling capacity. It is assumed that individual condenser pumps are dedicated to individual chillers and provide rated flow rates at rated power consumption. Chilled-water pump power per unit design capacity is not included in Equation 1 because it is relatively independent of chiller type, depending more strongly on the design of the chilled-water distribution system. The use of rating information to determine optimal chiller sequencing implies that it is best to allow operating chillers to reach their full cooling capacity before bringing an additional chiller online. This is generally the case for chillers that employ dedicated chilled and condenser water pumps. However, if two chillers have the same rated operating cost, then the chiller with the better part-load performance should have a higher priority.

At each decision interval (e.g., five minutes), the following steps can be executed to determine chiller sequencing lists that are used in bringing chillers online or offline to meet load requirements at minimum operating cost:

1. Use current utility energy rates, chiller performance at rating conditions, and typical chiller maintenance cost information to determine the cost of operating each individual chiller (i) per unit of cooling using Equation 1.

2. Use the chiller operating cost information determined in step 1 to construct a prioritized list for chiller sequencing (lowest to highest cost) for operating in the absence of demand limiting. This is the Stage 0 chiller sequencing list, [StageList.sub.0](i), i = 1, [N.sub.ch] .

3. Set the stage list counter to zero, k = 0.

4. Scan the kth-stage chiller sequencing list from the k+1 position to the end, i = k+1 to [N.sub.ch], and find the first location, m, that contains a gas-driven chiller. This is the next available demand-limiting chiller having the lowest operating cost.

5. Increment the stage list counter, k = k + 1.

6. The sequencing list for the kth stage of demand limiting is constructed by modifying the list from the previous stage (k?1). The next demand-limiting chiller determined in step 4 is placed in the kth position, bumping the remaining chillers down a position in the list. The following algorithm does the job:
If (k > 1) then
Do i = 1, k-1
 [StageList.sub.k](i) = [StageList.sub.k-1](i)
[StageList.sub.k](k) = [StageList.sub.k-1](m)
Do i = k, [N.sub.ch]
 If (i [not equal to] m) then
 [StageList.sub.k](i+1) = [StageList.sub.k-1](i)


7. If k is less than the number of gas-driven chillers ([N.sub.g]) then go to step 4. Otherwise, exit the algorithm.

Consider an example that employs the parameters specified in Table 1 with four different chillers and an on-peak period having electrical energy costs of $0.1/kWh and gas energy costs of $0.80/therm. Rated information for the chillers is summarized in Table 3.
Table 3. Rated Information for Example Chillers

No. Chiller Type COP Maintenance Condenser Condenser
 Costs, Water Pump
 $/ton-h Flow, Power,
 gpm/ton W/gpm

E-1 Variable-Speed 6.0 0.008 3 15
 Electric

E-2 Fixed-Speed 6.0 0.008 3 15
 Electric

G-1 Engine-Driven 1.5 0.016 3 15

G-2 Absorption 1.0 0.01 4 15



The first step in developing sequencing lists is to determine the operating cost per unit of cooling in the absence of any demand limiting. Equation 1 is used along with the example data and appropriate unit conversions. For the electric chillers (E-1 and E-2),

[j'.sub.ch, rated, E - 1,2] = [ER.sub.E - 1,2] * [1/[COP.sub.rated, E - 1,2]] + [MR.sub.E - 1,2] + [ER.sub.e] * [p'.sub.cwp, rated, E - 1,2] = 0.1[$/[kWh]] * ([1/6] * 3.517[kWh]/[ton-h]) + 0.008[$/[ton]] + 0.1[$/[kWh]]*(0.015[[kW]/[gpm]] * 3[gpm]/[ton]) = 0.07112$/ton-h

For the engine chiller (G-1),

[j'.sub.ch,rated,G - 1] = [ER.sub.[G - 1]] * [1/[COP.sub.[rated,G - 1]]] + [MR.sub.[G - 1]] + [ER.sub.e] * [p'.sub.[cwp.rated,G - 1]] = 0.8[$/[therm]] * (1/1.5 * 0.12[therm]/[ton-h]) + 0.016[$/[ton]] + 0.1[$/[kWh]] * (0.015[[kW]/[gpm]] * 3[gpm]/[ton]) = 0.0845$/ton-h

For the absorption chiller (G-2),

[j'.sub.[ch,rated,G - 2]] = [ER.sub.[G - 2]] * [1/[COP.sub.[rated,G - 2]]] + [MR.sub.[G - 2]] + [ER.sub.e] * [p'.sub.[cwp,rated,G - 2]] = 0.8[$/[therm]] * (1/1.0 * 0.12[therm]/[ton-h]) + 0.01[$/[ton]] + 0.1[$/[kWh]] * (0.015[[kW]/[gpm]] 4[gpm]/[ton]) = 0.0112$/ton-h

The Stage 0 sequencing list is established in Table 4 by ordering the chillers from lowest to highest operating cost per unit of cooling. Although the variable-and fixed-speed electric chillers have identical costs at the rating condition, the variable-speed chiller performs more efficiently at part-load conditions and would have lower overall operating costs. The electric chillers have significantly lower operating costs than the engine and absorption chiller for this example.
Table 4. Chiller Sequencing List for Stage 0 (No Demand Limiting)

No. Chiller Type [j'.sub.ch, rated], $/ton-h

E-1 Variable-Speed Electric 0.0711
E-2 Fixed-Speed Electric 0.0711
G-1 Engine-Driven 0.0845
G-2 Absorption 0.1120



In order to establish the Stage 1 demand-limiting sequencing list, the gas chiller in the Stage 0 list having the lowest operating cost is identified. In this case, it is the engine-driven chiller G-1. This chiller is placed at the top of the Stage 1 list and the other chillers are moved down. The resulting Stage 1 list is given in Table 5.
Table 5. Chiller Sequencing List for Stage 1 Demand Limiting

No. Chiller Type [j'.sub.ch, rated], ($/ton-h)

G-1 Engine-Driven 0.0845
E-1 Variable-Speed Electric 0.0711
E-2 Fixed-Speed Electric 0.0711
G-2 Absorption 0.1120



The Stage 2 demand-limiting list is established by moving the second most efficient gas chiller to the second position in the Stage 2 list and moving the other chillers down. The resulting Stage 2 list is given in Table 6. There are only two stages of demand limiting for this example because there are only two gas chillers.
Table 6. Chiller Sequencing List for Stage 2 Demand Limiting

No. Chiller Type [j.sub.ch, rated'], ($/ton-h)

G-1 Engine-Driven 0.0845
G-2 Absorption 0.1120
E-1 Variable-Speed Electric 0.0711
E-2 Fixed-Speed Electric 0.0711


Bringing Chillers Online of Offline

A chiller can be brought online for three reasons: (1) the current chillers can no longer meet the load (i.e., the chilled-water supply temperature set point cannot be maintained), (2) the plant power consumption is getting close to the plant target demand, or (3) the utility rates have changed (i.e., transition from offpeak to onpeak rates) and the current chiller sequencing is no longer the most economical. It is important to limit chiller transitions to avoid cost penalties associated with startup and shutdown of this large equipment. With these issues in mind, the following steps can be executed at each decision interval (e.g., five minutes) to bring chillers online or offline.

1. If the utility rate period has changed (e.g., offpeak to onpeak period) since the last control decision evaluation, go to step 2. Otherwise go to step 4.

2. If the target demand limit for the current rate period is not enabled (i.e., target set to an artificially high value) and the current chiller sequencing is different than what would be determined from the stage 0 chiller sequencing list, then shut down all chillers except the chillers that are at the top of the stage 0 list, set the demand-limiting stage to 0 ([Stage.sub.d] = 0), and exit the algorithm. Otherwise, go to step 3.

3. If the target demand limit is equal to zero and the current chiller sequencing is different than what would be determined from the stage [N.sub.g] (i.e., last) chiller demand-limiting list, then shut down all chillers except the chillers that are at the top of the stage [N.sub.g] list, set the demand-limiting stage to [N.sub.g] ([Stage.sub.d] = [N.sub.g]), and exit the algorithm. Otherwise, go to step 4

4. Evaluate the time-averaged value of the plant power consumption, chilled-water supply temperature, and overall cooling load over a fixed time interval (e.g., five minutes).

5. If individual chiller part-load ratios are not controllable and the current plant power consumption is greater than a specified fraction (e.g., 0.98) of the target demand for the current rate period and the current demand-limiting stage is less than the maximum number of stages ([Stage.sub.d] < [N.sub.g]), then increment the demand-limiting stage ([Stage.sub.d] = [Stage.sub.d] + 1).

6. If a feedback controller is utilized with part-load control of demand-limiting chillers to regulate plant power and the feedback controller output has been saturated at its maximum value (e.g., 1) for a fixed period of time (e.g., five minutes) and the current demand-limiting stage is less than the maximum number of stages ([Stage.sub.d] < Ng), then increment the demand-limiting stage ([Stage.sub.d] = [Stage.sub.d] + 1).

7. If individual chiller part-load ratios are not controllable and the current plant power consumption is less than a specified fraction (e.g., 0.80) of the target demand for the current rate period and the current demand-limiting stage is greater than zero ([Stage.sub.d] > 0), then decrement the demand-limiting stage ([Stage.sub.d] = [Stage.sub.d]-1).

8. If a feedback controller is utilized with part-load control of demand-limiting chillers to regulate plant power and the feedback controller output has been saturated at its minimum value (e.g., 0) for a fixed period of time (e.g., five minutes) and the current demand-limiting stage is greater than zero (Staged > 0), then decrement the demand-limiting stage ([Stage.sub.d] = [Stage.sub.d]-1).

9. Compare a list of current operating chillers with the list associated with the current demand-limiting stage. If the current operating chillers do not follow the priorities established by the current demand-limiting stage, then shut down the lowest priority chiller that has been operating for a minimum period of time (e.g., 30 minutes).

10. If the chilled-water supply temperature is significantly greater than the setpoint (e.g., 1[degrees]F) or there is another indication that a chiller or multiple chillers are operating near full cooling capacity, then bring the next chiller online from the list associated with the current demandlimiting stage and exit the algorithm. Otherwise, go to step 11.

11. If the cooling load is significantly less (e.g., 10%) than the load associated with the first time interval after the last chiller was brought online, then bring that chiller offline.

Update Chiller Loadings

Typically, chillers are arranged in parallel and have identical chilled-water setpoint temperatures. In this case, individual chiller loading can only be controlled by adjusting individual chilled-water flow rates using continuously adjustable valves. However, this is rarely employed. In the absence of individual flow control, the flows should be balanced so that each operating chiller has approximately the same partload ratio with the same chilled-water setpoints. This would ensure that each chiller could utilize its entire range of capacity control.

With individual flow control, there is an opportunity to adjust individual chiller loadings to reduce operating costs. In the absence of demand limiting (stage 0), the goal is to find the loadings that minimize the instantaneous energy costs. Braun (2007b) investigated different heuristic strategies for chiller loading and found that the cost penalties associated with applying equal part-load ratios (chiller load relative to chiller capacity) are relatively small. Braun (2007b) also developed an algorithm for determining chiller loads that is based upon minimizing cost, which results in somewhat better performance. However, the added benefits may not be sufficient to justify the added complexity.

It can be advantageous to adjust part-load ratios for demand-limiting chillers when controlling to a specific demand target. In this case, the best strategy involves using the demand-limiting chillers on the top of the chiller priority list to control the plant power demand to match the target using a feedback controller. The demand-limiting chillers are loaded with equal part-load ratios at a level determined by a feedback controller that attempts to keep the plant power consumption at the demand target specified by the overall cost optimizer. The remainder of the chiller priority list is treated in the same way as the chiller priority list for stage 0 (no demand limiting). This strategy minimizes the operating costs subject to the demand constraint.

Update Cooling Tower Fan Settings

Braun (2007a) presents development, evaluation, and implementation of a general algorithm for controlling cooling tower fans for hybrid cooling plants that was designed to minimize energy costs and can be employed in combination with chiller sequencing and loading strategies presented in this paper. The fan control strategy is separated into two parts: tower airflow and fan sequencing. Expressions are given for estimating the optimal tower airflow relative to maximum airflow that minimizes the combined cost of fans and chillers. Different expressions are given for different fan types and relate relative airflow to individual part-load ratios for chillers, gas and electric energy rates, and several other factors that are derived from design information for the chillers and cooling towers. Once the optimal tower airflow is established, then it is converted to a specific set of tower fans using optimal sequencing rules. Braun (2007a) presents detailed implementation steps and an example. 7Simulated plant cooling costs associated with the algorithm were compared with costs for optimized settings and were within 1% of the minimum costs.

Update Plant Demand Target

Braun (2007b) developed a simple criterion for determining monthly demand targets that gives a solution of either zero or infinite demand limit. A target limit of zero corresponds to a strategy of minimizing demand costs by maximizing use of the gas chillers during the rate period. This strategy makes sense in situations of high demand rates. A target limit of infinity corresponds to a strategy of minimizing energy costs. The best choice between these two limits depends on the energy cost penalty associated with maximizing the use of gas chillers as compared with the demand cost reduction. In turn, the results depend on utility rates for gas costs, electrical energy and electrical demand, the load profile, and plant configuration and performance.

Each utility period (e.g., on-peak, mid-peak, off-peak) should have its own demand target. For a given set of rates, the same target limits can be applied for all months where the rate period applies. The determination of the most appropriate demand target strategy for a rate period (minimum demand or minimum energy cost) may require that an analysis of the plant be performed. In some situations, the solution is obvious. For instance, the demand target should be set to infinity for offpeak periods that do not have demand charges. Conversely, for on-peak periods in summer, the demand target would normally be set to zero. Otherwise, the overall economics of employing gas chillers would be suspect. However, for other cases the result may have a significant dependence on utility costs, load profile, and plant characteristics and would require that an analysis be performed for a representative month (e.g., July for summer season utility rates) to evaluate which limiting case should be applied.

For utility periods that should utilize an energy cost minimization strategy (e.g., no off-peak demand charges), then demand limiting should be disabled by setting an artificially high value for the demand target. For utility periods that should utilize a demand minimization strategy, then the demand target should be set to 0 at the start of each billing period (i.e., each month). For subsequent days in the month, the minimum demand cost is continuously updated within the rate period as the minimum of the current demand and the previous target. This approach reduces energy cost penalties for the minimum demand strategy. Typically, demand charges are levied based on the peak average demand occurring over a fixed moving window in time (e.g., 15 minutes). With these issues in mind, the demand target for each utility rate period is updated at each control decision interval (e.g., five minutes) using the following steps.

1. If the billing period has changed since the last time that the algorithm was executed for the current rate period, then reset the demand target to 0 for minimum demand control or to an artificially large number for minimum energy cost control and exit the algorithm. Otherwise go to step 2.

2. Evaluate the time-averaged plant power demand that will be used for levying utility demand charges (e.g., 15minute average).

3. Update the plant demand for the current rate period as the maximum of the current plant power demand and the previous demand target using

[P.sub.p1,limit,p,k] = Maximum{[P.sub.p1,k,][P.sub.p1.limit,p,k-1]]}

DEMAND-LIMITING CONTROL BEHAVIOR

The simplified sequencing and loading strategies were implemented within the cooling plant model along with simplified tower control and used to study the effects of plant load and target demand on plant energy costs and electrical demand. Figure 3 shows results for a four-chiller plant consisting of all the chiller types considered in this study. The plots show plant energy costs per unit of cooling, total plant electrical demand, and chiller sequencing as a function of plant part-load ratio for the design ambient condition. Each of the chillers will handle a quarter of the plant load. Without a demand limit in place, the chiller sequencing follows the energy priority list given in Table 4. For a given combination of chillers operating, the cost per unit of cooling generally decreases with loading. Also, the costs increase at the points where additional chillers are brought online. The electrical demand generally increases with load but decreases at the points where gas chillers are brought online.

[FIGURE 3 OMITTED]

Figure 4 shows similar plots for a case where a demand limit was set at 50% of the electrical demand for the plant operating at capacity. Two cases are considered: (1) no part-load control of individual chillers and (2) part-load control of demand-limiting chillers (equal part-load ratios for remaining chillers). The results are the same as for no demand limit until the plant load is about 30% of capacity. After this point, chillers are sequenced as required according to the demand-limiting stages specified in Tables 5 and 6. For this case, the sequencing strategy is able to keep the demand below the target (~500 kW) until about 75% of plant capacity. Part-load control results in lower demand above this capacity and slightly lower plant costs for some of the lower plant loads. The same chiller sequencing is used for both cases, but part-load control allows the demand limit to be maintained near the target over a wider range of loads.

[FIGURE 4 OMITTED]

Figure 5 shows results for a demand limit set at 25% of the electrical demand for the plant operating at capacity (~250 kW). In this case, the differences between part-load and no-partload control are greater. Between about 25% and 40% of the plant capacity, part-load control of the gas chillers resulted in significantly lower energy costs because it allowed the demand target to be maintained with a more efficient combination of chillers. Above 50% of plant capacity, the target demand could not be maintained, but part-load control resulted in significantly lower demand for a range of loads.

[FIGURE 5 OMITTED]

There are some cases in the results of Figures 4 and 5 where the part-load control resulted in higher energy costs because the strategy is designed to control the cooling plant power demand below a target. However, higher energy costs do not necessary mean higher utility costs. If the target is chosen appropriately, then the monthly utility costs (including energy and demand cost) are lower for part-load control as compared with no-part-load control.

SAVINGS POTENTIAL FOR DEMAND-LIMITING CONTROL

The cost savings associated with employing demand-limiting control for hybrid cooling plants can be very significant, depending on a variety of factors including utility rates and load and plant characteristics. Figure 6 gives an overall snapshot of the annual cost savings potential for all combinations of systems described in Table 2. The annual costs per ton-hour of cooling for demand-limiting control are plotted versus costs associated with setting all demand limits to infinity, meaning that chiller sequencing and loading are based on minimizing energy costs only. In all cases, the near-optimal tower control strategy was employed and chillers could be loaded at uneven part-load ratios when in demand-limiting mode. The costs with demand-limiting control can be as little as 50% of the energy cost optimal results.

[FIGURE 6 OMITTED]

The cost savings for demand-limiting control generally increase with increasing ratio of demand to total utility costs. Figure 7 plots annual percent savings for demand-limiting control as a function of the ratio of demand to total plant costs for no demand limiting for all of the simulations considered for Figure 6. Although there is a strong correlation, other factors, such as load profile and plant configuration, play a role. The savings for demand limiting control can still be significant when demand costs are between 10% and 20% of the total utility costs.

[FIGURE 7 OMITTED]

Figures 8 and 9 give some specific example results that show how the savings for demand limiting depend on demand charges, gas rates, building type, and location. Generally, greater cost savings occur with higher electrical demand charges and lower gas costs. For the office in Houston, the savings are relatively small. The savings are substantially higher for the school in San Francisco. Compared to the office, the school has a relatively short occupancy period with a peak load that is large relative to the integrated load. The milder climate in San Francisco also leads to a higher ratio of peak to integrated load. As a result, the school in San Francisco has a much higher ratio of demand costs to energy costs than the office in Houston. This leads to substantially greater opportunities for demand reduction through the use of hybrid cooling plants. In general, the use of hybrid cooling plants with demand-limiting control is more beneficial for buildings with shorter occupancy periods having high demand charges.

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

Figure 10 shows annual demand-limiting cost savings comparisons for cases with and without part-load control of demand-limiting chillers. The ability to control the load on demand-limiting chillers to achieve a target plant demand target leads to additional savings compared to just changing the chiller sequencing. However, there are still significant savings without part-load control under appropriate circumstances.

[FIGURE 10 OMITTED]

CONCLUSIONS

The control strategies presented in this paper for hybrid cooling plants complement the strategies that are presented for all-electric plants within the 2003 ASHRAE Handbook--HVAC Applications on "Supervisory Control Strategies and Optimization" (chapter 41). Application of the demand-limiting strategy could lead to very significant savings, particularly for buildings having high ratios of demand to total costs. In general, the use of hybrid cooling plants with demand-limiting control is more beneficial for buildings with shorter occupancy periods having high demand charges.

ACKNOWLEDGMENTS

The financial support of ASHRAE under research project RP-1200 and the technical support provided by the Project Monitoring Subcommittee, which included Jay Kohler, Paul Sarkisian, and Dharam Punwani, are greatly appreciated.

NOMENCLATURE

[COP.sub.rated,i] = rated chiller COP for the ith chiller

[ER.sub.e] = cost per unit of electrical energy

[ER.sub.i] = cost per unit energy input (heat or electricity) for the ith chiller (gas or electric)

[j'.sub.ch,rated,i] = operating cost per unit time and design cooling capacity

[MR.sub.i] = maintenance cost per unit of runtime and rated capacity for the ith chiller (gas and electric)

[N.sub.ch] = total number of chillers (gas and electric)

[N.sub.g] = number of gas-driven chillers within the cooling plant

[p'.sub.cwp, rated, i] = condenser water pump power per unit of rated chiller cooling capacity for ith chiller

[P.sub.pl] = total power associated with the chiller plant

[P.sub.pl,limit] = peak plant power consumption target limit

REFERENCES

Arnold, R.L., and W.P. Bahnfleth. 1998. Peak shaving using natural gas engine-driven chillers. Heating, Piping, Air Conditioning 70(9):51-59.

ASHRAE. 2003. 2003 ASHRAE Handbook--HVAC Applications, Chapter 41, Supervisory control strategies and optimization, pp. 41.1-41.39. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Braun, J.E. 2006. Optimized operation of chiller equipment in hybrid machinery rooms and associated operating and control strategies. ASHRAE RP-1200 final report, American Society of Heating, Refrigerating, and AirConditioning Engineers, Atlanta, GA.

Braun, J.E. 2007. A general control algorithm for cooling towers in cooling plants with electric and/or gas-driven chillers. HVAC&R Research 13(4):581-98.

Braun, J.E. 2007. Near-optimal control strategies for hybrid cooling plants. HVAC&R Research 13(4):599-622.

GTI. 2005. Natural gas-fired cooling technologies and economics. Available at www.gastechnology.org/webroot/app/xn/xd.aspx?it= enweb&xd=GasCooling\ home.xml&bd=demo. Gas Technology Institute

Meckler, M. 1997. Case for hybrid gas/electric chiller plants. Energy Engineering: Journal of the Association of Energy Engineering 94(5):73-77.

Nowakowski, G.A., and M. Gramlich. 1999. New developments in hybrid natural gas/electric cooling. ASHRAE Journal 41(9):36-42.

NREL. 1995. User's Manual for TMY2 (Typical Meteorological Years), NREL/SP-463-7668, and TMY2s, Typical Meteorological Years Derived from the 1961- 1990 National Solar Radiation Data Base, June 1995, CD-ROM. Golden, CO: National Renewable Energy Laboratory.

Smith, B. 2002. Economic analysis of hybrid chiller plants. ASHRAE Journal 44(7):42-46.

Wylie, M., and R. Alvarez. 1997. 'Hybrid' central plants. Energy Engineering: Journal of the Association of Energy Engineering 94(2):6-16.

James E. Braun, PhD, PE

Fellow ASHRAE

James E. Braun is professor of mechanical engineering at Purdue University, West Lafayette, Indiana.
COPYRIGHT 2007 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2007 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Braun, James E.
Publication:ASHRAE Transactions
Article Type:Report
Geographic Code:1USA
Date:Jul 1, 2007
Words:5870
Previous Article:100 ton absorption chiller/heat pump demonstrates the real cost of saving energy.
Next Article:Exergetic performance analysis of Various cogeneration systems for buildings.
Topics:


Related Articles
Simulated performance of natural and hybrid ventilation systems in an office building.
Jaeggi/Guentner.
General methodology combining engineering optimization of primary HVAC & R plants with decision analysis methods--Part I: deterministic analysis.
General methodology combining engineering optimization of primary HVAC & R plants with decision analysis methods--Part II: uncertainty and decision...
A near-optimal control strategy for cool storage systems with dynamic electric rates (RP-1252).
A General control algorithm for cooling towers in cooling plants with electric and/or gas-driven chillers.
Near-optimal control strategies for hybrid cooling plants.
Enhancement and use of combined simulation tools in the assessment of hybrid natural/mechanical ventilation systems.
Impact of control on operating costs for cool storage systems with dynamic electric rates.
Supervisory and optimal control of building HVAC systems: a review.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters