BACKGROUND OF THE INVENTION 0001 The present invention relates generally to controlling temperature of a fluid in a refrigeration system. More specifically, the present invention relates to using fuzzy logic to control the rate at which the leaving chilled water temperature of a chiller is pulled down to a predetermined set point while minimizing overshoot of the predetermined setpoint. 0002 In a chilled water system, chilled water is circulated through a building to remove heat from the building. The water in the chilled water system is cooled in an evaporator or chiller using a heat exchanger, wherein the water is cooled by a refrigerant that accepts heat from the water. Chilled water systems are often operated, for optimum efficiency, at temperatures (operating setpoints) within a few degrees of the freezing point of water. In existing chillers, a control system is often programmed to shut down the chiller as soon as the evaporator water temperature decreases to a certain temperature (a cutout point) near or below the freezing point of water to avoid freezing the water tubes and damaging the chiller. For example, a chiller may have an operating setpoint of 35 F. and a cutout point of 34 F. 0003 A difference of only one degree between the operating setpoint and the cutout point in the chiller, generally does not cause problems during normal or steady-state operation of the chiller. However, when the chiller is required during a pull-down to reduce the water temperature from an ambient temperature to the operating setpoint, a one degree difference between the operating setpoint and the cutout point can be problematic. The chiller may be shut-down by the control system during a pull-down, if the water temperature in the chiller overshoots the operating setpoint and reaches the cutout point. To avoid this problem, it is necessary that chiller control systems be programmed to minimize the pull-down overshoot without sacrificing the pull-down response time. 0004 In standard proportional-integral-derivative (PID) controls, the derivative term controls the rate of approach to the temperature setpoint and is manually tuned and then fixed to achieve a desired maximum overshoot/undershoot for worst case conditions. However, as system dynamics change due to load variations, etc., the response time may become too slow under certain conditions due to the fixed derivative (rate) term. One way to avoid the problems associated with the fixed derivative term is to use a self-tuning control which automatically adjusts the rate term to maintain the desired response time while keeping overshoot/undershoot within acceptable limits. These self-tuning algorithms require knowledge of the amount of undershoot/overshoot before making appropriate adjustments to compensate the control. However, these algorithms may result in the chiller shutting down during the initial and subsequent pull-downs, since the control has not yet completely compensated for the initial overshoots. 0005 Therefore, what is needed is a fuzzy logic control algorithm that can maintain a desired pull-down response time and minimize or eliminate overshoot of an operating setpoint by adjusting a rate sensitivity coefficient of the control algorithm during a mode of pull-down operation that is commenced when the water temperature reaches a few degrees above the operating setpoint.

SUMMARY OF THE INVENTION 0006 One embodiment of the present invention is directed to a temperature control system for a refrigeration system. The refrigeration system comprises a compressor, a condenser, and an evaporator all connected in a closed refrigeration circuit. The compressor has a plurality of inlet guides vanes that are adjustable and controlled by an actuator. The temperature control system includes a sensor for detecting the temperature of the leaving fluid in the evaporator and for providing a temperature signal. The temperature control system also includes a microprocessor that samples the temperature signal at a predetermined interval during a pull-down operation on the leaving fluid temperature in the evaporator. The microprocessor then generates a control signal for the actuator by applying the temperature signal to a fuzzy logic control algorithm configured to minimize or eliminate overshoot of a setpoint temperature of the leaving fluid temperature of the evaporator during the pull-down operation. 0007 Another embodiment of the present invention is directed to a method of calculating a control signal to control flow of refrigerant in a refrigeration system during an operation to adjust the temperature of a secondary refrigerant in the refrigeration system to a predetermined setpoint temperature. The method includes the step of generating a temperature signal for the secondary refrigerant in the refrigeration system. An error signal is generated using the temperature signal and a rate signal is generated using the error signal. A control signal is generated using fuzzy logic reasoning in response to the error signal, the rate signal and a rate sensitivity signal to control the flow of refrigerant in the refrigeration system. Finally, the flow of refrigerant in the refrigeration system is controlled to adjust the temperature of the secondary refrigerant in the refrigeration system to minimize or eliminate overshoot of the predetermined setpoint temperature in response to the control signal. 0008 Yet another embodiment of the present invention is directed to a method of calculating a control signal for an actuator of a plurality of vanes to control flow of refrigerant from an evaporator during a pull-down operation to adjust a temperature of leaving water from the evaporator to a predetermined setpoint temperature. The method comprises the steps of measuring, at predetermined intervals, a temperature of the leaving water from the evaporator. An error value is calculated from the difference between the measured temperature and the predetermined setpoint temperature and a pull-down rate value is calculated from a difference between the error value and a previously calculated error value from the prior interval. Next, a control signal is generated at each predetermined interval using a fuzzy logic control algorithm in response to the error value, the pull-down rate value and an adjustable rate sensitivity value. Finally, the actuator is operated in response to the control signal to position the plurality of vanes to control flow of refrigerant from the evaporator to minimize or eliminate overshoot of the predetermined setpoint temperature during adjustment of the temperature of the leaving water from the evaporator to the predetermined setpoint temperature. 0009 Still another embodiment of the present invention is directed to a refrigeration system comprising a compressor, a condenser, and an evaporator connected in a closed refrigeration circuit. A refrigerant is circulated through the refrigeration circuit. Furthermore, the compressor has a plurality of inlet guides vanes adjustable by an actuator. The refrigeration system also includes a sensor for detecting a leaving fluid temperature from the evaporator to provide a temperature signal and a microprocessor generating an error signal, an error rate of change signal and an adjustable rate sensitivity signal at a predetermined interval during a pull-down operation on the leaving fluid temperature in the evaporator in response to receiving the temperature signal. The microprocessor also generates a control signal for the actuator of the plurality of inlet guide vanes using a fuzzy logic control algorithm. The fuzzy logic control algorithm has the error signal, the error rate of change signal and the adjustable rate sensitivity signal as inputs. The control signal is used to control refrigerant flow from the evaporator to minimize or eliminate overshoot of a setpoint temperature for the leaving fluid temperature of the evaporator during the pull-down operation. 0010 A further embodiment of the present invention is directed to computer program product embodied on a computer readable medium and executable by a microprocessor for calculating a control signal for an actuator of a plurality of vanes to control flow of refrigerant from an evaporator during a pull-down operation to adjust a temperature of leaving water from the evaporator to a predetermined setpoint temperature. The computer program product comprising computer instructions for executing the steps of measuring, at predetermined intervals, a temperature of the leaving water from the evaporator. The computer program product also comprises computer instructions for executing the steps of calculating an error value from the difference between the measured temperature and the predetermined setpoint temperature, calculating a rate value from a difference between the error value and a previously calculated error value from a prior interval, and calculating an adjustable rate sensitivity value using a fuzzy logic algorithm. In addition, the computer program product comprises computer instructions for executing the step of generating a control signal at each predetermined interval using a fuzzy logic control algorithm in response to the error value, the rate value and the adjustable rate sensitivity value. The control signal, when transmitted to the actuator, operates the actuator to position the plurality of vanes to control flow of refrigerant from the evaporator to minimize or eliminate overshoot of the predetermined setpoint temperature during adjustment of the temperature of the leaving water from the evaporator to the predetermined setpoint temperature. 0011 One advantage of the present invention is that a control algorithm is used to minimize or eliminate pull-down overshoot without sacrificing response time. 0012 Other features and advantages of the present invention will be apparent from the following more detailed description of the preferred embodiment, taken in conjunction with the accompanying drawings which illustrate, by way of example, the principles of the invention.

BRIEF DESCRIPTION OF THE DRAWINGS 0013 FIG. 1 illustrates schematically a refrigeration system of the present invention. 0014 FIGS. 2 and 3 illustrate graphs of membership functions of inputs to the fuzzy logic control algorithm. 0015 FIG. 4 illustrates a fuzzy logic truth table of the fuzzy logic control algorithm. 0016 FIG. 5 illustrates a fuzzy logic truth table from one example of the present invention. 0017 FIG. 6 illustrates a graph of the membership function for a fuzzy logic system for determining rate sensitivity during a pull-down mode of operation. 0018 FIGS. 7A and 7B illustrate a flow chart of the process for determining the rate sensitivity during the pull-down mode of operation. 0019 Whenever possible, the same reference numbers will be used throughout the figures to refer to the same parts.

DETAILED DESCRIPTION OF THE INVENTION 0020 A general system to which the invention is applied is illustrated, by means of example, in FIG. 1. As shown, the HVAC refrigeration system 100 includes a compressor 110, a condenser 112, a water chiller or evaporator 126, and a control panel 140. The control panel 140 includes an analog to digital (A/D) converter 148, a microprocessor 150, a non-volatile memory 144, and an interface board 146. The operation of the control panel 140 will be discussed in greater detail below. The conventional liquid chiller system includes many other features which are not shown in FIG. 1. These features have been purposely omitted to simplify the drawing for ease of illustration. 0021 The compressor 110 compresses a refrigerant vapor and delivers it to the condenser 112. The compressor 110 is preferably a centrifugal compressor, however other types of compressors can be used such as scroll compressors, reciprocating compressors, and screw compressors. The condenser 112 includes a heat-exchanger coil 116 connected to a cooling tower 122. The condensed liquid refrigerant from condenser 112 flows to an evaporator 126. The evaporator 126 includes, for example, a heat-exchanger coil 128 having a supply line 128S and a return line 128R connected to a cooling load 130. Water or any other suitable secondary refrigerant, e.g. ethylene, calcium chloride brine or sodium chloride brine, travels into the evaporator 126 via return line 128R and exits the evaporator 126 via supply line 128S. The evaporator 126 chills the temperature of the water in the tubes. The heat-exchanger coil 128 can include a plurality of tube bundles. The vapor refrigerant in the evaporator 126 then returns to compressor 110 via a suction line to complete the cycle. At the input to the compressor 110 from the evaporator 126, there are one or more pre-rotation vanes or inlet guide vanes that control the flow of refrigerant to the compressor 110. An actuator 120 is used to open the pre-rotation vanes to increase the amount of refrigerant to the compressor 110 and thereby increase the cooling capacity of the system 100. Similarly, the actuator 120 is used to close the pre-rotation vanes to decrease the amount of refrigerant to the compressor 110 and thereby decrease the cooling capacity of the system 100. 0022 The system 100 includes a sensor 160 for sensing the temperature within the evaporator 126 that corresponds to the leaving chilled water temperature (LCHWT) from the evaporator 126. The sensor 160 is preferably in the refrigerant flow, at the outlet pipe or supply line 128S from the evaporator shell. However, the sensor 160 can be placed in any location that provides an accurate measurement of the LCHWT. The sensor 160 is preferably a temperature thermistor, however, other types of temperature sensors may also be employed. The thermistor provides a resistance that is proportional to the temperature. The resistance from the thermistor is then converted to a voltage signal, using a resistor divider connected to a voltage source or any other suitable technique for generating a voltage. The voltage signal from the sensor 160 is transferred over a line 162 to the control panel 140. 0023 The voltage signal input to control panel 140 over line 162 is converted to a digital signal or word by A/D converter 148. This digital signal representing the evaporator temperature can now be converted by the microprocessor 150 into a corresponding LCHWT. The LCHWT value is then input into the control algorithm, which is described in more detail in the following paragraphs, to generate a control signal for the pre-rotation vane actuator 120. The control signal for the pre-rotation vane actuator 120 is provided to the interface board 146 of the control panel 140. The interface board 146 then provides the control signal to the pre-rotation vane actuator 120 to position the pre-rotation vanes. 0024 Microprocessor 150 uses a fuzzy logic algorithm to control the pre-rotation vane actuator 120 through the interface board 146. In one embodiment, the fuzzy logic algorithm can be a computer program having a series of instructions executable by the microprocessor 150. The control algorithm determines during a pull-down of the LCHWT, whether to cause the pre-rotation vane actuator 120 to further open the pre-rotation vanes, further close the pre-rotation vanes, or maintain the pre-rotation vanes in the same position, thereby regulating the flow of refrigerant vapor and correspondingly the rate that the LCHWT approaches the predetermined operating setpoint temperature. 0025 While it is preferred that the fuzzy logic control algorithm is embodied in a computer program and executed by the microprocessor 150, it is to be understood that the fuzzy logic control algorithm may be implemented and executed using digital and/or analog hardware by those skilled in the art. If hardware is used to execute the fuzzy logic control algorithm, the corresponding configuration of the control panel 140 can be changed to incorporate the necessary components and to remove any components that may no longer be required, e.g. the A/D converter 148. 0026 The fuzzy logic algorithm controls the desired position of the pre-rotation vanes by periodically sampling the LCHWT measured by the sensor 160 at pre-programmed intervals. For example, the programmable intervals may range from 6 to 12 seconds. By sampling the output of the sensor 160 and comparing the sensed values with a predetermined operating setpoint temperature, and with one or more previously stored samples of the LCHWT, the microprocessor 150 can calculate an error value (en) and the rate of change of the error (den), according to known computer techniques. The ultimate goal of the fuzzy logic algorithm is to make the error value approach zero so that the operating setpoint temperature is achieved with little or no overshoot that approaches the cut-out point to cause the chiller 126 to shut-down. During each sample interval, the fuzzy logic algorithm of the microprocessor 150 determines the degree of negative, positive and zero membership associated with each input (the error (en) and its rate of change (den)) by assigning a weight between zero and one to each input. Then, the fuzzy logic algorithm evaluates several if then rules that combine the degrees of membership into the appropriate course of action for the pre-rotation vane actuator 120. 0027 The fuzzy logic algorithm utilizes as variable inputs both a LCHWT error (en) and the LCHWT error rate of change (den). The LCHWT error (en) is calculated as the LCHWT from the sensor 160 minus the predetermined operating setpoint temperature and the LCHWT error rate of change (den) is the quantity of the present LCHWT error (en) minus the LCHWT error from the previous sample (enl) multiplied by the rate sensitivity. The rate sensitivity is a programmable value which ranges from 50-220 and is preferably 140 during steady-state operation. The determination of the rate sensitivity will be provided in greater detail below. 0028 In a preferred embodiment, the fuzzy logic algorithm includes five membership functions for the LCHWT error rate of change (den) and three membership functions for the LCHWT error (en). FIG. 2 illustrates the membership functions for the LCHWT error rate of change (den) and FIG. 3 illustrates the membership functions for the LCHWT error (en). Each membership function determines in a linear fashion the degree to which the given input is zero, positive or negative. Full membership corresponds to a value of 1.0, partial membership corresponds to a value between 0 and 1.0 and no membership corresponds to a value of 0. The LCHWT error (en) membership functions are described by three linguistic variables Negative Small (NS), Zero (ZR) and Positive Small (PS). The LCHWT error rate of change (den) membership functions also include the linguistic variables Negative Large (NL) and Positive Large (PL) in addition to the three linguistic variables Negative Small (NS), Zero (ZR) and Positive Small (PS). 0029 The membership functions shown in FIGS. 2 and 3 are symmetric about zero and reflect the same degree of membership for negative values as positive values when considering inputs of equal magnitude. In general, membership functions may or may not be symmetric. The membership functions are independently programmable and may be changed in the microprocessor 150. Thus, the sensitivity of both the LCHWT error (en) and the LCH VWT error rate of change (den) membership functions may be modified, both symmetrically and asymmetrically, as desired, to optimize the system control. It is preferred to have programmable membership functions in order to have the flexibility to tune the control. A user can then change the membership functions applied by the fuzzy logic algorithm to obtain a desired system response. However, the membership functions may also be predetermined and non-programmable to provide the user with a simpler and easier to use control arrangement. 0030 The table shown in FIG. 4 shows a fuzzy logic truth table which diagrams how the microprocessor 150 evaluates the fuzzy logic rules according to one embodiment of the present invention. The fuzzy logic rules can be evaluated to one of five output linguistic variables Close Large (CL), Close Small (CS), No Response (NR), Open Small (OS) and Open Large (OL). The output linguistic variables in the truth table are shown with numbered subscripts to distinguish them for illustration purposes. 0031 As shown in the table, the rule combinations, described in the format of de n membership function, en membership function, are as follows: (Negative Large (NL), Negative Small (NS)), (Negative Small (NS), Negative Small (NS)), and (Negative Large (NL), Zero (ZR)) result in a Close Large (CL) output variable and contribute to a close command for the pre-rotation vane actuator 120 to reduce the cooling capacity of the chiller 126; (Negative Large (NL), Positive Small (PS)), (Negative Small (NS), Zero (ZR)), and (Zero (ZR), Negative Small (NS)) result in a Close Small (CS) output variable and contribute to a close command for the prerotation vane actuator 120 to reduce the cooling capacity of the chiller 126; (Positive Large (PL), Zero (ZR)), (Positive Small (PS), Positive Small (PS)), and (Positive Large (PL), Positive Small (PS)) result in a Open Large (OL) output variable and contribute to a open command for the pre-rotation vane actuator 120 to increase the cooling capacity of the chiller 126; and (Positive Small (PS), Zero (ZR)), (Zero (ZR), Positive Small (PS)), and (Positive Large (PL), Negative Small (NS)) result in a Open Small (OS) output variable and contribute to a open command for the pre-rotation vane actuator 120 to increase the cooling capacity of the chiller 126. The remaining three rule combinations are not evaluated since they result in no action or no response. Therefore, a total of twelve rule combinations are evaluated utilizing a fuzzy inference minimum/maximum method. This method implies that a minimum fuzzy AND inferencing will be performed first for each of the twelve rule combinations. Then a fuzzy OR inferencing will be performed in which the maximum value is found for the three rule combinations that result in the Close Large (CL) output variable contribution and similarly for the Close Small (CS) output variable contribution, as well as for the three rule combinations that result in the Open Large (OL) output variable contribution and similarly for the Open Small (OS) output variable contribution, thus resulting in four maximum values representing resultant open and closed values. The equations for determining the contribution of each output variable are provided below.

CLMAX (CL1, CL2, CL3)(1)

CSMAX (CS1, CS2, CS3)(2)

OSMAX (OS1, OS2, OS3)(3)

OLMAX (OL1, OL2, OL3)(4)

0032 The resulting four maximum values need to be combined into a single output response, i.e., they need to be de-fuzzified. Since the centroid method of defuizzification is more computationally intensive than required for this application, the singleton approach is preferably used. In the singleton approach, each of the output contributions, i.e. the maximum value of the Close Large (CL) output variable, the Close Small (CS) output variable, the Open Large (OL) output variable and the Open Small (OS) output variable, are each multiplied by a weighting or scaling factor. Preferably, the scaling factor is 6 for the Close Large (CL) output variable and the Open Large (OL) output variable and 3.5 for the Close Small (CS) output variable and the Open Small (OS) output variable, although different scaling factors can be used. Next, the single output response is determined by subtracting the scaled close contribution maximum values from the scaled open contribution maximum values and then multiplying the subtraction result by a pulse factor. The equation for determining the output response or output pulse width (OPW) is provided below.

OPW(b 6*OL)(3.5*OS)(3.5*CS)(6*CL)* Pulse_Factor(5)

0033 The Pulse_Factor multiplier is a programmable value that may range from 0.3 to 1.0. The selection of the Pulse_Factor multiplier permits a user to further tune the system response by enabling a pulsewidth adjustment to the output response. A larger value for the Pulse_Factor multiplier results in a more sensitive system response and a smaller value for the Pulse_Factor multiplier results in a less sensitive system response. 0034 If the resultant value is less than zero, the pre-rotation vane actuator 120 receives a pulse width on the close signal (negative value) of equal value in seconds to the resultant value for that sample period. If the resultant value is positive, the pre-rotation vane actuator 120 receives a pulse width on the open signal (positive value) of equal value in seconds to the resultant value for that sample period. An error deadband of approximately 0.2 F. results in no response to either the open or close signal. 0035 The fuzzy inferencing of the present invention will be further clarified by the following example, which is intended to be purely exemplary of the invention. For this example, the LCHWT error (en) is 1 F., which, recall from above, is calculated as the LCHWT from the sensor 160 minus the predetermined operating setpoint temperature, and the LCHWT error rate of change (den) is 5 F./sec., which, recall from above, is the quantity of the present LCHWT error (en) minus the LCHWT error from the previous sample (enl) multiplied by the rate sensitivity. As shown in FIG. 2, the LCHWT error rate of change (den) of 5 F./sec. yields degrees of membership of 0.5 to NL and 0.5 to NS. As shown in FIG. 3, the LCHWT error (en) of 1 F. yields degrees of membership of 0.75 to ZR and 0.25 to PS. The rules are combined by the minimum method and shown in the fuzzy logic truth table of FIG. 5. A comparison is made between the top and left side membership values of the truth table and the minimum value is placed in its respective place in the output section of the truth table. For example, the degree of membership assigned to the NL membership function of den (0.5) is combined with the degree of membership assigned to the PS membership function of en (0.25) by performing a minimum fuzzy inferencing, i.e., a fuzzy AND routine, and results in a minimum value of 0.25 being assigned to CS1. Next, the results of the combinations or fuzzy AND routines are combined using maximum fuzzy inferencing is performed, i.e., a fuzzy OR routine, as provided below in equations (6)-(9), which results in CS0.5, CL0.5, OS0 and OL0.

CLMAX (0, 0, 0.5)0.5(6)

CSMAX (0.25, 0.5, 0)0.5(7)

OSMAX (0, 0, 0)0(8)

OLMAX (0, 0, 0)0(9)

0036 These results are then defuzzified using equation (5) discussed above and a Pulse_Factor of 1 to obtain an output response or pulse width of 4.75. This output results in a 4.75 second pulse width on the close signal being executed during the current period. In other words, the output response is calculated and executed during the current sample interval and before the next sampling event. 0037 To achieve a smooth approach to the operating setpoint with little or no overshoot, the rate sensitivity term is adjusted during a temperature pull-down situation. The rate sensitivity term is adjusted based on the pull-down rate (dTenenl) at which the LCHWT is approaching the predetermined operating setpoint and the difference between the predetermined operating setpoint and the LCHWT or the LCHWT error (en). When the LCHWT equals or is greater than a preselected value (preferably 6 F., but other values can be used) above the predetermined operating setpoint, i.e. the LCHWT error (en) equals or is greater than the preselected value, the control algorithm operates in a pull-down mode of operation instead of in a steadystate mode of operation. In the pull-down mode of operation, the sampling period is automatically set to the minimum sampling period, which is preferably 6 seconds, and the rate sensitivity is calculated based on the pull-down LCHWT rate (dT). The pull-down or self-tuning mode of operation is preferably concluded when the LCHWT error (en) becomes negative and then becomes positive again or two minutes after the LCHWT has reached a value that is within 0.5 F. of the predetermined operating setpoint, i.e. the LCHWT error (en) <0.5, however, other determinations for concluding the pull-down mode can be used. When the pull-down mode of operation is concluded the rate sensitivity and the sampling period are automatically returned to their normal programmed values for steady state control operation. 0038 In the temperature pull-down mode of operation, the rate sensitivity term is determined using fuzzy reasoning. FIG. 6 illustrates the membership functions for the pull-down LCHWT rate (dT). The pull-down LCHWT rate (dT) membership functions are described by two linguistic variables, Negative (NEG) and Positive (POS). Recall from above, that when the pull-down mode is entered the sampling period is set to 6 seconds, thus the values of dT illustrated in FIG. 6 are based on a sample period of 6 seconds, i.e. dT is calculated in degrees Fahrenheit per 6 seconds ( F./(6 sec.)). The membership functions for the pull-down LCHWT rate (dT) are dependent on a target rate, which defines the point that yields 0% membership for both membership functions. The membership functions for the pull-down LCHWT rate (dT) extend from the target rate to upper and lower extremes that yield 100% membership. The target rate is programmable and adjustable, as described in greater detail below. If a pull-down LCHWT rate (dT) is obtained that is less than the target rate (i.e. a negative value), the rate sensitivity will be increased because the pull-down LCHWT rate (dT) is moving too fast and has to be slowed down. In contrast, if a pull-down LCHWT rate (dT) is obtained that is greater than the target rate, the rate sensitivity will be decreased because the pull-down LCHWT rate (dT) is moving too slow and has to be increased. 0039 In a preferred embodiment of the present invention, the target rate is fixed in the pull-down mode at 0.3 F./(6 sec.) or 0.05 F./sec. when the LCHWT error (en) is greater than 2 F., however other fixed target rates can be used for this temperature range. After the LCHWT error (en) reaches 2 F. or less in the pull-down mode, the target rate is decreased linearly as the LCHWT error (en) approaches zero to slow the rate of approach so that minimal to zero overshoot of the predetermined setpoint is achieved, however, other function types for decreasing the target rate can also be used. FIG. 6 illustrates the membership functions for the pull-down LCHWT rate (dT) for a LCHWT error (en) greater than 2 F. (solid lines) and for a LCHWT error (en) between 0 and 2 F. (dashed lines). 0040 The fuzzy inferencing of the present invention will be further clarified by another example, which is intended to be purely exemplary of the invention. For this example, the pull-down LCHWT rate (dT) is 0.4 F./(6 sec.), the LCHWT error (en) is greater than 2 F. and the target rate is 0.3 F./(6 sec.) As shown in FIG. 6, the pull-down LCHWT rate (dT) of 0.4 F./(6 sec.) yields degrees of membership of 0.33 to NEG and 0.0 to POS. The rate sensitivity is then determined from the following equation provided below.

rate_sensitivity140(NEG*110)(POS*90)(10)

0041 Solving equation (10) with degrees of membership of 0.33 to NEG and 0.0 to POS yields a rate sensitivity of 176 as shown below in equation (11).

rate_sensitivity140(0.33*110)(0*90)176(11)

0042 FIGS. 7A and 7B illustrate a flow chart of the process for determining the rate sensitivity in the pull-down mode of operation. The process begins in step 1002, where the LCHWT error (en) is checked to see if it is greater than or equal to 2 F. If the LCHWT error (en) is greater than or equal to 2 F. then the target rate is set to 0.3 F./(6 sec.) in step 1004. If the LCHWT error (en) is less than 2 F., then the LCHWT error (en) is checked to see if it is greater than or equal to 0 F. in step 1006. If the LCHWT error (en) is greater than or equal to 0 F., then the target rate is determined in step 1008 using the equation provided below.

target_rate0.3(((2.0error (en))*0.3)/2.0)(12)

0043 If the LCHWT error (en) is less than 0 F., then the target rate is set to 0 F./(6 sec.) in step 1010. Next, after the target rate has been determined in steps 1004, 1008, 1010, the negative rate limit is determined in step 1012 using the equation provided below.

neg_rate_lim0.3target_rate(13)

0044 In step 1014, the pull-down LCHWT rate (dT) is checked to see if it is less than or equal to the negative rate limit. If the pull-down LCHWT rate (dT) is less than the negative rate limit, then the degree of membership for the Negative membership function is set to 1.0 and the degree of membership for the Positive membership function is set to 0.0 in step 1016. In step 1018, the pull-down LCHWT rate (dT) is checked to see if it is less than or equal to the target rate. If the pull-down LCHWT rate (dT) is less than or equal to the target rate, then the degree of membership of the Negative membership function is determined in step 1020 using the equation provided below.

NEG(rate (dT)target_rate) (neg_rate_limtarget_rate)(14)

0045 Also in step 1020, the degree of membership for the Positive membership function is set to 0.0. In step 1022, the pull-down LCHWT rate (dT) is checked to see if it is greater than the positive rate limit. If the pull-down LCIIWT rate (dT) is greater than the positive rate limit, then the degree of membership for the Positive membership function is set to 1.0 and the degree of membership for the Negative membership function is set to 0.0 in step 1024. If the pull-down LCHWT rate (dT) is less than the positive rate limit, then the degree of membership of the Positive membership function is determined in step 1026 using the equation provided below.

POS(rate (dT)target_rate)/(pos_rate_limtarget_rate)(15)

0046 Also in step 1026, the degree of membership for the Negative membership function is set to 0.0. Finally, after the degrees of membership have been determined in steps 1016, 1020, 1024 and 1026, the rate sensitivity is calculated in step 1028 using equation (10) as discussed above. The calculated rate sensitivity is then used in the fuzzy logic methodology described above to calculate a control signal for the pre-rotation vane actuator 120. 0047 The present invention has been described in the context of a control algorithm for a pre-rotation vane actuator that controls the flow of refrigerant from an evaporator and to a compressor in a chiller system, thereby controlling the pull-down rate of the leaving chilled water temperature of the evaporator. However, the control algorithm of the present invention can be used in any type of refrigeration system to control the rate, either pull-down or pull-up, at which the temperature of a fluid in the system is brought to a setpoint. To use the control algorithm in other types of refrigeration systems, some changes may have to be made to the membership functions and the sensor information that is used by the control algorithm to account for the particular configuration of the system to which the control algorithm is being applied. 0048 While the invention has been described with reference to a preferred embodiment, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the appended claims.