Adaptive control is a proven, traditional control procedure that is at work on many diverse control applications. Adaptive control monitors actual system conditions and adjusts the control algorithm to fit those conditions. It can adjust points of transition of equipment such as chillers, cooling towers, and pumps.

Digital electronics has enabled a control engineer to apply adaptive control to central chilled water plants. Now, adaptive control can monitor actual conditions that exist in and around a chiller plant and adjust equipment operation to acquire the highest possible COP or lowest kW/ton.

No longer is it necessary to program equipment on the calculations that were made for the selection of chillers, towers, and pumps. Setpoints that have been inserted into the program manually during commissioning no longer need to be continually adjusted by the operators to achieve efficient operation. Adaptive control can do this through specific control procedures used for chilled water pumps, condenser pumps, low return-water temperature, and for the chiller plant in general.

Programming of Chilled Water Pumps

Most chilled water pumps of any size are now variable speed. Their speed is usually controlled by system conditions such as the differential pressure that exists at the end of the main loops of the chilled water system. New control technologies are coming into existence that may offer other means of controlling pump speed.

The other aspect of variable-speed pump control is programming pumps on and off. All larger chilled water systems have two or more chilled water pumps operating in parallel to achieve greater efficiency and to provide standby capability. The points at which standby pumps are started and stopped are achieved by a number of means. Recent evaluation of this subject indicates that the most economical procedure from an instrumentation point is the use of the total kW power input to the pumping system.

Total kW input to a pumping system can be calculated at all loads on the chilled water system, from minimum to maximum flow. These calculations are completed with any possible number of pumps running such as one, two, three, or four pumps in operation on a four-pump system. The results are kW input curves as are shown in Figure 1.

These curves are generated from the design engineer's data, which are the result of careful evaluation before design is initiated. It is difficult for the engineer to determine the maximum chilled water flow required due to the diversity of the cooling load and actual population of the space being cooled. Added to this is the inexactness of friction calculations for pipe and fittings along with the need to provide for aging of the piping and equipment. Further, the design criteria may include provisions for future cooling loads. All of this results in the actual operating conditions being quite different than the design conditions.

Figure 1 describes the kW input to a four-pump system with a capacity of 4,500 gpm at a pump head of 100 ft. This data was based on the design criteria for a chilled water system. As indicated in this figure, the kW input demonstrates that one pump should operate up to 1,250 gpm, two pumps to 2,300 gpm, and three pumps up to 3,250 gpm even though each pump has a design capacity of 1,500 gpm. These curves demonstrate the need for energy analysis when operating pumps in parallel.

Figure 1 describes the performance based on design conditions while Figure 2 includes the actual operation. As is often the case, the maximum flow is less than the design flow; it was 4,000 gpm, not 4,500 gpm. In view of the difference between calculated and actual system conditions, one pump should now operate up to 1,450 gpm, two pumps up to 2,650 gpm, and all four pumps above 3,650 gpm.

Operating conditions may vary from these, which would require changing the transition points for the pumps to achieve minimum energy consumption. This has been the practice in the past with the operator observing the points of transition and adjusting them manually to secure the optimum number of pumps in operation.

With adaptive control, the pump controller can recognize increases in energy consumption and automatically adjust the point of transition to always have the most efficient number of pumps in operation. When a pump is added or subtracted, the energy consumption should drop; otherwise, there would be no reason for the change in the number of pumps in operation. This is so since any number of pumps is always operating below the maximum capacity required to serve the chilled water system.

If the energy consumption rises during the transition, the controller recognizes a need for change in the point of transition, and when this point next occurs, the transition point will be adjusted. This process is operated continuously in the controller software.

Condenser Pumps

Condenser pumps may or may not be variable speed, depending on the needs of the chillers. Some chiller plants do not require variable-speed condenser pumps, while others do. If the condenser pumps are variable speed, the same control technique described above for the chilled water pumps can be applied to the condenser pumps.

Sequencing Chillers

Adding and subtracting chillers with chilled water system load changes has been a subject open for discussion, particularly with return chilled water coming to the chillers at temperatures below the design temperature. Generally, the control procedure for chiller addition is to use leaving water as the control parameter. The running chillers should be loaded as heavily as possible to achieve a lower energy consumption per ton of cooling. Figure 31 describes the kW/ton for a 750-ton chiller operating with a constant condenser flow of 2,250 gpm at an entering temperature of 75 degrees F at all percentage loads on the chiller.

These values were kept constant to demonstrate changes in energy consumption due to different loadings on the chiller. The curves in Figure 3 are 1) for the chiller alone and 2) with a constant speed cooling tower fan and condenser pump. Both the cooling tower fan and condenser pump were equipped with 40-hp motors. The cooling tower fan motor was fully loaded, while the condenser pump motor was operating at 33 hp. The total kW input for the cooling tower and condenser pump was a constant 59 kW.

It is obvious that the steepness of the curve in Figure 3 for the chiller, cooling tower, and condenser pump necessitates that the chillers operate at as high of loadings as possible. The full-load energy use of the chiller compressor was 435 kW, but, with the condenser water at 75 degrees, this power was only 385 kW or 88.5% of design. If it would be possible to overload the running chiller to 110% flow, the actual chiller kW would be 431 kW, still below the design power of 435 kW.

If a chiller were added when the running machine reached design flow, the two chillers would be running at 50% load or 0.68 kW/ton, including the power for the condenser pump and cooling tower. When the second chiller is now added, each would be running at 55% load and 0.65 kW. Although this represents a power savings of only around 5%, the important fact to remember is that with adaptive control, the chillers can be added at the most efficient point, regardless of the chilled water temperature or flow.

Adaptive control would recognize the actual conditions that exist and add the standby chiller when certain system conditions existed, such as a power supply of 435 kW, rising leaving water temperature above the setpoint, or maximum evaporator velocity of 12 fps. It would remember the chiller flow or tonnage at the point of addition and would stop the standby chiller when the system conditions dropped below the point of chiller addition. Fixed points of addition and subtraction of chillers will no longer be needed. This is particularly so when the return water temperature from the chilled water system is below the design temperature.

Low Return-water Temperature

Low return water temperature from a chilled water system is known in the industry as "low return water temperature syndrome." It has earned this nickname due to the disastrous results that it can have on chiller performance. The basic reason for applying adaptive control to chiller sequencing is to overcome as much as possible the effects of low water temperature on chillers.

In actual practice, it is not uncommon to find the return water temperature to be six to eight degrees below the design water temperature. For example, a number of chiller plants have been designed with 44 degrees supply and 56 degrees return water temperature. In actual operation, the return water temperature can sink to as low as 48 degrees. Also, the return water temperature is not constant; it may vary with the cooling load on the chilled water system. An example of the effect on chiller performance is following with actual return water temperatures as low as 48 degrees.

At 48 degrees return water temperature with 44 degrees supply water temperature, the differential temperature is 4 degrees or 6 gpm /ton of cooling. At 1,500 gpm, the design flow for the chiller, the chiller will be producing only 250 tons of cooling, not 750 tons. If another chiller is added at this point, each will be producing only 125 tons of cooling. If a chiller is to be stopped when the waterflow drops to 1,300 gpm, each chiller will be producing 108 tons of cooling or 14% of full load. At this point just before the second chiller is stopped, each chiller by itself would have an energy input of 0.83 kW/ton.

Adaptive control can be applied to chiller sequencing by increasing the flow in the evaporator until one of three things occur: 1) erosion velocity in the evaporator tubes is approached, 2) the leaving temperature from the chiller starts to rise, or 3) the rated power for the chiller, in kW, is approached. Each of these conditions will result in the addition of another chiller. Since return water temperature to a chiller plant is a variable, there can be no setpoint at which the next chiller is added.

When a chiller is added, the adaptive control records the chiller plant conditions and stops the standby chiller at a point that will not restart the standby chiller immediately. This is demonstrated as follows:

If the lead chiller is allowed to run alone until 2,250 gpm is flowing in the evaporator, it will be generating 375 tons of cooling before the second chiller is started. By holding off the addition of another chiller until the maximum output is achieved from the running chiller, the energy savings will be as high as 28% of that with two chillers operating.

If two chillers are run at 2,250 gpm and the second chiller is stopped when the plant conditions ensure that one chiller can handle the cooling load, each chiller will be generating 175 tons of cooling just before the chiller is stopped. At this point, each chiller will be producing 23% of design load and will have an energy input of 0.65 kW/ton. The reduction in energy by using adaptive control for starting and stopping chillers is 0.65/0.83 or 22%. Obviously, these are calculated savings, but the adverse effect of low return water temperature on chiller operation has been observed many times in the field.

Clearly, the principal need here is to increase the return water temperature, but this is caused by conditions outside of the chiller plant. Normally, dirty cooling coils, oversized or ineffective cooling coil control valves, or thermostats set too low are the causes for low return water temperature. Water bypassing in a primary/secondary pumping system will also lower the return water temperature.

Adaptive control applied to chiller sequencing can eliminate some of the effects of low return water temperature. As indicated above, the return water temperature in an actual chiller plant does not remain at one temperature, and adaptive control can automatically adjust the chiller add and subtract points to the existing return chilled water temperature. In summation, adding and subtracting chillers can be made much more efficient through the use of adaptive control.

Overall Chiller Plant Performance

Considerable work is being done to improve overall chilled water plant performance. Not only have chillers increased appreciably in efficiency with higher COPs or lower kW/ton, but also digital control is now available to operate the entire chiller plant with lower energy consumptions. The principal question now is: How do you orchestrate the operation of the chillers, cooling towers, and condenser pumps to achieve the lowest total energy consumption at any cooling load on the chiller plant?

Initially, instrumentation is needed that will determine with reasonable accuracy the cooling load that exists on the chilled water plant. ASHRAE has established the GPC-22 committee for the determination of a "Guideline for the Determination of the In Situ Coefficient of Performance for Electric Motor Driven Central Chilled Water Plants."

The proposed instrumentation for the generation of the actual COP or kW/ton for the electric motor-driven chilled water plant is shown in Figure 4. A significant specification for the instrumentation will be that all instrumentation must have its accuracy traceable to NIST (National Institute of Standards and Technology). The objective here is to achieve an overall instrument accuracy of around ± 1% to 2% throughout a flow range of 1 to 15 ft/sec.

If this COP or kW/ton for the total chilled water plant is made available to the plant operators, they will be able to compare present operation with past performance. They will be able to determine the correct number of chillers, cooling towers, and pumps that produces the minimum consumption of energy. Also, this should be an operational tool that will assist them in establishing maintenance schedules. With adaptive control, much of the manual control of setpoints will be eliminated from their work.

So where does adaptive control assist in the improvement of overall chiller plant operation? Along with improving the sequencing of pumps and chillers, it can do iterations that will determine the optimum condenser flow rates and temperatures. For example:

  • Assuming constant leaving water temperature from the cooling tower, the rate of condenser waterflow can be varied incrementally by adaptive control until the maximum COP or minimum kW/ton is achieved for the total plant.
  • wise, the condenser flow can be kept constant, and the approach temperature of the cooling tower can be narrowed or widened to reach optimum energy conditions.
  • Variations of these two iterations may be practical for some chiller plants. Not all chiller plants will need any of these operations, while others will be candidates for them.

So who can do adaptive control, and what does it cost? Adaptive control is a procedure that is included in almost any textbook on control, so anyone can use it. Its cost is dependent on the control equipment in place for a chiller plant. The software for adaptive control may be added to some existing chiller plant controls. It may require a new controller that may cost in the range of $20,000 to $30,000. Obviously, the energy savings possibilities for a particular installation will determine the economic feasibility of its installation.

It should be emphasized that the savings in energy included herein are for very specific parameters. These actual figures are unimportant. What is important is that we now have software capabilities for evaluating energy savings for central chilled water plants, particularly those operating with low return chilled water temperatures.ES

The images associated with this article do not transfer to the Internet. To review the figures, please refer to the print version of this issue.

1 The data for these curves was derived from calculations furnished by York International Corporation.