Control sequences for chilled water systems

The most important component of good chiller plant control is understanding the full and part load requirements and then matching those to system performance.

By Peter D. Zak, PE, Graef-USA Inc., Milwaukee January 20, 2012

As the headline implies, this article will provide a discussion and overview of methods used to control chillers. To understand where we are today, a little time travel is required.

The year 1902 is significant because that is when Willis Haviland Carrier designed a “temperature and humidity” conditioning system for a printing plant in Brooklyn, N.Y. Temperature and humidity fluctuations created problems when workers tried to align multiple ink colors in the printing process.

In 1911 Carrier submitted to the American Society of Mechanical Engineers his basic “rational psychometric formulae,” which to this day is considered the basis of all calculations with respect to the properties of air.

Carrier received a patent for the first centrifugal refrigeration machine in 1921. The design was inspired by observing the construction and operating characteristics of a centrifugal pump. At the time, the centrifugal compressor was considered a more practical and safe method of providing air conditioning for large spaces in lieu of the traditional reciprocating units that used ammonia, CO2, or other gases considered hazardous.

In 1924 the Hudson Department Store in Detroit became the first commercial building to have an air conditioning system specifically designed for human comfort. Until that point most air conditioning (cooling) was part of an industrial process. Thus was born “human comfort cooling,” thanks to the acknowledged father of air conditioning, Willis Haviland Carrier.

Moving forward to the present, we find the same basic principles apply, but the delivery method is being constantly revised. Mechanical components, such as compressor types and heat transfer surfaces, have changed in response to changing refrigerants and maximizing unit efficiencies. When there was cheap energy and the public was just thankful to be out of the heat and humidity, simple on/off control methods were perfectly acceptable. As long as the space was cooler than the outdoors, a 5 or 10 F temperature swing was just fine.

Back then and even now, the on/off concept is simple and works, although it does have a detrimental effect on operational life of the equipment. Rising energy costs and the demand for more precise temperature control are the driving factors for properly and efficiently controlling a chiller plant. In an ideal world, the cooling required and subsequent energy use would exactly match. In reality, this never happens due to system efficiencies, controllability, and the thermodynamic time lag that occurs in buildings.

The whole system

The challenge for design engineers is just how to maximize the comfort requirements while keeping energy costs to a minimum. This is best achieved by not only properly controlling each individual component of a chiller plant but also effectively controlling the entire system.

Accurately determining the chiller load is the first step. It is important not just to have a grasp of the maximum and minimum load requirements, but also to understand the building load profile that identifies the time duration during part load events. This critical information is needed to identify the quantity and desired performance requirements of the equipment.

A typical chiller plant consists of a chiller or multiple chillers, along with chilled water pumps, condenser water pumps, and cooling towers. The selection of this equipment is driven by the individual efficiencies at full and partial loads, and also by how the system efficiency will respond to the building loads. The individual cumulative efficiencies of the components are not a realistic value.

Each piece of equipment in a chiller plant has different part load efficiency. The trick is to know what those values are, when that load occurs for that specific component, and how to synchronize that equipment’s operating points to maximize plant efficiency.

In the chiller, the main system components include a compressor or something that moves the refrigerant; an evaporator, which is a heat exchanger that transfers heat from the chilled water to the refrigerant; and condensers, additional heat exchangers that transfer the heat picked up from the chilled water to the refrigerant and the heat of compression to either the air or a water heat rejection sink.

The design output capacity of a chiller exists only for a short period of time during the cooling season. A large percentage of operating hours are spent at some point between off and 100% loaded. To compensate for this wide range of loading conditions, control devices were added that will modulate or restrict the flow of refrigerant in response to the setpoint of the chiller leaving water temperature. This is accomplished by unloading cylinders in the reciprocating unit, modulating a slide valve in a screw-type compressor, or controlling refrigerant flow with a variable speed drive or inlet vanes on a centrifugal compressor. The objectives are to maintain the unit operation during part load conditions, avoid on/off cycles, and minimize energy consumption.

If you were to plot the operation of the various types of chiller compressors, you would see that the reciprocating unit will unload in steps. This means that for more precise load matching requirements, more steps are required.

A simple analogy would be trying to fit rectangular objects under a curve—the larger and fewer the objects, the more open area under or above the curve. The smaller and more frequent the rectangles, the closer you get to match the curve. The gaps under the curve indicate unmet loads and unit inefficiency. Areas over the curve indicate unnecessary energy consumption and overcooling.

The curve actually represents the part load performance of the chiller. This may be a shock to some of the hardcore digital techies, but the load curve is analog. A compressor that can have the refrigerant flow controlled infinitely will be more successful matching the loads and maintaining efficiency.


Chillers are rated by the American Refrigeration Institute (ARI). Part load performance is based on a fixed leaving chilled water temperature and fluctuating condenser water temperature in response to the outdoor air temperature.

From the machine perspective, the objective of the chiller is to make cold water. How efficiently this is accomplished is based on the type of compressor and how the internal control system modulates the refrigerant flow. This becomes a little more complicated when multiple machines are used. This is just one component of the total efficiency of a system.

On the chilled water side, most pumping configurations use variable flow strategies. The two most common configurations are a single pump with standby or pumps in parallel. There are pros and cons for each configuration, but the proper selection will include an analysis of the operational characteristics based on the hourly operational loads. This will provide a profile of the power requirements and efficiencies. This information will indicate the most efficient operating point for the maximum hours of system load.

A single pumping system is simple because a single system curve is followed; a parallel pumping system becomes a little more complex. Each pump is selected at 50% of total system flow. The curves are additive and the system curve is plotted through both pump curves. As the system load changes, the flow requirements change as well as system head. Horsepower changes exponentially to the third power in relationship to flow. Pump speed is controlled by differential pressure transmitters located throughout the chilled water system.


There is a misconception that in a parallel pumping configuration, once the system flow drops to  anything less than design flow rate, one pump must be de-energized. In reality, it may be more efficient to operate both pumps at a reduced capacity, based on overall efficiency. Control logic used to determine this is called wire to water efficiency. A system operational curve is programmed into the control logic. Pressure sensors and flowmeters calculate the system horsepower. Current transducers on the pump motors measure the actual power consumption. Data is compared to the programmed system curve that then selects the most efficient number of pumps to be operated. This holds true for any number of pumps operating in a parallel configuration.

Condenser water pumps can be configured in the same manner. There are several control strategies for condenser water pump control. The most simple is constant flow with three-way mixing valves at each condenser. The three-way mixing valves modulate in response to refrigerant head pressure or a temperature sensor set to maintain a fixed entering or leaving condenser water temperature. Good design practice will have a control valve for each chiller. In plants with a mixture of chiller sizes and types, it is imperative that each condenser have an independent control valve.

In variable flow pumping strategies, a two-way valve modulates condenser water flow in response to refrigerant head pressure. Differential pressure transducers control the pump speed. In some scenarios the pumps can be controlled by the head pressure; however, this becomes difficult with multiple chillers and requires a very sophisticated control system.

Maximum chiller efficiency is achieved by controlling refrigerant flow and getting the coldest possible water to the condensers. Most chiller manufacturers will accept chilled water as low as 65 F. For example, in centrifugal chillers, temperatures below this limit will adversely affect the lubricating system and control of refrigerant flow. This cold water is produced at the cooling towers, which makes the selection and operation of the towers an integral component to maximizing the overall system efficiency.

The most common cooling towers are cross-flow type. Warm condenser water is sprayed over a media or fill material and air is moved through the material via a fan. The evaporative properties of air temperature and moisture cool the condenser water. Cooling tower capacity and performance is based on the outdoor wet bulb temperature and the ability of the equipment to match the wet bulb temperature and the tower leaving water temperature, which is known as the approach temperature. The approach temperature is the value given to the differential temperature between two media. In this example if the outdoor wet bulb temperature is 78 F and the condensing water temperature leaving the tower is 85 F the approach temperature would be 7 F. As mentioned earlier, chillers work well with cold condenser water; the ARI ratings and chiller performance curves indicate that unit efficiencies increase during part load cycles with colder condenser water.

When selecting a tower, it may be cost-effective to maximize the fill media within a nominal unit casing. The more fill, the more surface area—and the better the heat transfer.


Tower controls can be as simple as an on/off fan based on a fixed leaving water temperature, or two-speed motors or variable frequency drives (VFD) to maintain a fixed setpoint. When towers are operated during freezing conditions, there must be a control algorithm that maintains a minimum flow through the tower to prevent ice buildup on the fill material.

If multiple cell towers are used, controlling the condenser water flow to each tower will allow all the tower surface media area to be used prior to starting the fans, which is an efficient method of operation. Once again, minimum flow rates and basin or sump volume are critical, as is maintaining heat transfer net positive suction head (NPSH), and preventing icing in cold weather use. While on/off control is simple, a major drawback will be the unstable water temperature. This is a critical function of chiller efficiency. A stable condenser water temperature will result in consistent chiller operation and efficiency.

Another added control feature is to have the leaving water temperature reset based on the outdoor air wet bulb temperature. This may require the tower fans to operate longer. The offset would be higher efficiency at the chillers, resulting in a net energy reduction for the system. Once again, the engineering evaluation must consider the impact of operating bypass valves or tower fans. Change only one thing; too many variables will never work.

So, how does all of this work in one cohesive unit? As previously discussed, each component has a primary function operating within its own design parameters and individual efficiencies. To operate as a cohesive energy-efficient system, there are individual component trade-offs that have net positive results.

To accomplish this, the controls or BAS has to be capable of monitoring and processing multiple data points that provide direction to the individual equipment components resulting in the most efficient operating configuration. This also includes the physical layout and configuration of the plant. Proper equipment placement based on functional relationships and correct water flow eliminate flow and pressure drop issues.

This all begins with either single or multiple sensors, located in a space or controlling a process temperature. Using a room thermostat as a starting point, the following events would occur:

  • As the space temperature increases above setpoint, a variable air volume damper modulates open. The fan speed increases in response to the duct static pressure controller. As more air is moved across the cooling coil, a discharge controller begins to modulate the two-way chilled water valve open.
  • As the chilled water valve begins to open, the chilled water differential pressure transmitter senses a drop in system pressure. This sends a signal to the chilled water control panel and VFDs. The pumps are designed to operate in a parallel configuration. Currently a single pump is operating in response to water flow requirements.
  • The request for additional water flow is processed prior to adjusting the pump speed. The system differential pressure sensor and water flow meter data are processed along with actual power consumption taken from the current transducer monitoring motor power requirements. The system wire to water efficiency is calculated and evaluated to determine if one or two pumps operating is appropriate prior to increasing the pump speed.
  • As the chilled water flow is increased, the additional load is noted at the chiller by an increased return water temperature. The chiller then begins to increase the flow of refrigerant by either adding more compressors in stages or changing the guide vanes or speed of a centrifugal chiller.
  • As this occurs, the total system water flow and chilled water temperature difference is monitored along with the chiller power consumption. This data is collected and processed by the BAS Calculated cooling tons, and unit kW/ton (part load) data are then used to optimize the staging of single or multiple chillers.
  • As the chiller loads modulate, a refrigerant head pressure controller modulates a two-way automatic valve on the condenser water side. Differential pressure transmitters sense the change in water pressure and control the condenser water pumps in the same manner as the chilled water pumps.
  • As additional water flow is presented to the cooling towers, an outdoor wet bulb sensor is used to reset the leaving water temperature controller. This device then opens the tower cell supply valves in sequence to each cell, first maximizing the use of fill material surface area in convective heat transfer prior to engaging the fans in sequence. Most of this works in reverse if the loads are being diminished.

So here we have an abbreviated outline of interoperability, demonstrating that a change in any one component will have an effect on overall system performance. As software becomes more sophisticated, the control sequences/operation will have an intuitive nature, preventing rapid system fluctuations due to a single data point change.

Controlling the system

Within the framework of the basic control strategy, there are a number of control subsets that provide equipment operating limits and safety controls. These control functions are essential to safely operate the equipment within the design parameters and guidelines of the manufacturer. Observing a typical wiring diagram, you will notice a configuration sometimes identified as a ladder diagram. This drawing indicates sensing devices specifically designed to prevent the operation of the unit unless a preset number of conditions are met. The circuits are wired in series, which means all conditions must be satisfied prior to allowing the unit to operate. For chillers, some of the more common items are high head pressure, low suction temperature, proof of water flow at the condenser and evaporator, oil pressure/temperature, starting current, and voltage.

For pumps, a proof of flow switch or differential pressure sensor across the pump is an accurate method to indicate pump operation. The sensing points should be directly across the pump suction and discharge. Sensing pressure after the pump discharge check valve will not work, especially in a multiple-pump common header design. Some designers use a current transducer on the motor that indicates the pump is running. However, if the pump uses a coupling to attach the pump shaft to the motor shaft, and that shears, the motor runs but the pump is not operational. The BAS may not recognize the failure and critical flow-sensitive equipment may also fail.

Cooling towers incorporate vibration sensors in the event a fan goes out of balance. Ice detection sensors protect the towers in cold climate operations as well as initiate a de-icing operation. Safety controls are usually inherent to the manufacturer and should not be altered. The BAS will monitor the safety/alarm circuits and indicate the situation at the user screen or, in more critical situations, initiate a visual or audible alarm.

Chilled water safety controls might include high- and low-temperature alarms for the chilled water and condenser water. Flow indications include minimum and maximum alarm limits and on/off equipment failure. In variable flow configurations, consideration must be given to evaporator or condenser minimum flow requirements.

The designer has total flexibility in identifying sensing points, alarm circuits, and control strategies. While this may bring joy to a control contractor, sometimes too much information will make system control difficult.

A chiller plant cannot respond instantaneously to a change in the cooling requirements. The relationship between time and load changes is different than the instantaneous world of electronic sensors and computers, so the sensitivity and response time must be reconciled.

Another common mistake is trying to reset a control point with too many variables. Pick one and only one. For example, do not control tower fan speed from chiller head pressure while trying to reset the leaving tower temperature based on wet bulb temperature.

There is a reason unit manufacturers have unit control panels for the equipment. They designed the equipment, they understand the operational limits, and they carry the warranty. Embrace this concept and integrate the BAS to work with the manufacturer’s unit controls. Do not reinvent the wheel; keep things simple. Most manufacturers’ software has the ability to integrate with the control system via BACnet, LonTalk, or Modbus. Attempting to bid individual control systems may prove difficult and may eliminate any equipment manufacturer’s warranty.

The most important component of good chiller plant control is understanding the full and part load requirements and then matching them to system performance.

There is a certain irony in the design and control of a chiller plant. In the beginning, the process was dedicated to satisfying the needs of an industrial process. Soon building and human comfort became the norm. As time went on, the tolerance for a wide offset in temperature with respect to setpoint weakened. This, along with rising energy costs, created a need for more efficient equipment along with precise accurate control.

The public’s developing awareness of operating costs, sustainable options, and the potential environmental issues of refrigerants is spawning a movement to accept larger temperature and humidity deviations. This is supported to some degree by the fact that an increase in natural (outdoor) air flow is becoming acceptable in the commercial environment, thus reducing the need for large chiller plants. To some extent we are moving back to the ways of 1921.

What would Willis Haviland Carrier think now?

Zak is a principal with Graef-USA Inc., headquartered in Milwaukee, where he manages the MEP group. He is a member of NCEES and is on the editorial advisory board of Consulting-Specifying Engineer. He was an adjunct assistant professor at the Milwaukee School of Engineering for 20 years and is a registered professional engineer in 24 states.