Historically, plant engineers have kept chiller operating logs to measure chiller performance and determine causes of problems. The data collected includes readings taken from the chiller during scheduled inspections such as evaporator and condenser temperatures, pressures, flows, running load amps, volts, etc. This schedule varies from every two hours to once a shift, depending on the type of operation and more importantly, manpower constraints. In most facilities, logs were a vital tool in scheduling downtime, preventive maintenance, and inspections based on chiller run hours. Today, it is common for facilities to maintain logs, but they rarely get reviewed until there is a problem, which is too late.

Garbage In

The old adage of "garbage in equals garbage out" holds especially true when determining chiller performance. All temperature and pressure sensors/gauges lose their calibration and drift over time. The period of time and amount of drift vary from one sensor to another. If your sensors haven't been calibrated in the last year and are over three years old, the odds of them being inaccurate are almost 100 percent. Evaporator and condenser water flows can fluctuate during seasonal changes, when circulating pumps begin to wear out or there are plant changes from original design. Therefore, if flows aren't measured and adjusted as needed, they are probably far from design.

Add the potential for human error into the equation. When logging data, it can be very easy to flip-flop readings, use outlet voltage instead of inlet voltage, transpose values, or use approximations when using mercury thermometers or dial gauges that are not properly sized for the operating pressures. Digital gauges that measure to the 1/10 are highly recommended for accuracy and the reduction of human error.

In most facilities today, manpower is a major problem. Cutting back on preventive maintenance and taking logs has become a target for saving manpower resources. It's hard to argue with an operating engineer when they say "I don't have time to take logs on a regular basis, much less evaluate the data." Compound this with inaccurate data, which renders it almost worthless and makes any clear-cut analysis virtually impossible. One can only hope that when problems occur they are not major, affecting operations and requiring expensive repairs.

Figure 1. EffHVAC daily report.

Determining Chiller Efficiency

A common method for determining chiller efficiency has been to calculate the actual kW/ton, then determine the difference between actual and design full-load kW/ton. The problem with this method is it can only be accurate if the chiller is operating at full load conditions when compared to design full load. This occurs on average less than 2 percent of the time. To calculate kW, take the square root of 3 (1.732), multiply by the actual running load amps, multiply by the actual volts, divide by 1,000, then multiply by the power factor (PF).

kW = [(1.732 x amps x volts) ÷ 1,000] x PF

To calculate tonnage, take the evaporator delta temperature (∆T), which is the difference between the evaporator water temperature in and the evaporator water temperature out, multiply by the evaporator water flow gallons per minute (gpm), and divide by 24.

Tons = (∆T x gpm) ÷ 24

Efficiency Technologies Inc. has taken these calculations and developed an Internet-based efficiency and trending tool for chillers, boilers, and plate exchangers called EffHVACâ„¢. EffHVAC accurately measures chiller performance at all part loads and any operating condition with a proprietary Calculated Part Load Value (CPLV). Comprehensive reports (Figure 1) include advanced diagnostics and fault detection that identify the cause of inefficiency (including bad data) and provide detailed corrective action instructions.

Determining Accuracy

There are some obvious ways to tell if your sensors are out of calibration. If the ∆T's vary high or low from design operating conditions, then it could be an indication that one or both of the temperature sensors are inaccurate or water flow may be off-design. If the evaporator leaving refrigerant temperature is greater than the evaporator chilled water out temperature, or the condenser leaving refrigerant temperature is less than the condenser water out temperature, then it indicates a problem in either the water temperature sensor or the refrigerant pressure sensor. If leaving refrigerant temperatures are not recorded, then pressures can be converted to temperature from a standard refrigerant/temperature chart, which can be downloaded at www.efftec.com. Without a tool like EffHVAC, it can be difficult to detect bad sensors and determine how far they are out of calibration.

Figure 2. Ice bath temperature measurement.

Garbage Out

The impact of inaccurate data can have a dramatic effect on energy consumption. Every 1°F decrease in chill water temperature caused by an inaccurate high reading creates a 2-4 percent increase in energy usage to maintain that unnecessary low temperature. Not knowing the real temperatures can cost a fortune in wasted energy, not to mention wear and tear on the chiller components by running outside of intended parameters. The four main contributors of bad data are temperature sensors, pressure sensors, water flow, and human error.

How do inaccurate temperature sensors/gauges affect efficiency calculations? Take an example of a chiller with design specifications of: design tonnage = 600, full load amps = 500, volts = 460, PF = 0.9, design ∆T = 10°F, gpm = 1,440 and design kW/ton = 0.598. If this chiller's evaporator water in temperature sensor is reading low 1°F and the evaporator water out temperature sensor is reading high 1°F, this gives a combined total of 2°F off or an 8°F ∆T at full load. When the kW/ton is calculated, it equals 0.747. Divide 0.598 by 0.747 to get 80 percent efficiency. Just 2°F drift can make it appear that the chiller is 20 percent inefficient.

This can alter scheduling of maintenance, produce inaccurate cost analysis, and skew the plant load profile by 20 percent, making decisions concerning chiller sizing very difficult. At 80 percent efficiency, a 600 ton chiller running at 50 percent load, 24 hours/365 days, at $0.06/kWh would indicate a $24,912 loss. This emphasizes why sensors and gauges must be accurately calibrated to 1/10°F.

Figure 3. Calibrating the temperature sensor.
Temperature Calibration
Temperature gauge calibration is very easy. Make sure the gauges to be calibrated are clean and dirt free. If a commercial temperature bath to perform a 3- or 5-point calibration is unavailable, freeze about 2 liters of deionized (DI) water and crush or shave into pieces as small as possible. Fill an insulated container, such as a Dewer flask or thermos that will maintain a temperature with the crushed ice. Fill the container with chilled DI water, and then drain the water out. Make sure the ice settles in the bottom of the container, minimizing air pockets in the ice. Continue this process until the volume is filled with ice and minimal liquid/air space. Submerge the entire length of the temperature probe into the ice bath until the readings stabilize (Figure 2). Adjust calibration dial until it reads 32.0°F (Figure 3). If calibrating more gauges, use the same ice bath, which can last for 1-2 hours.

Temperature sensors for a chiller panel or building automation system (BAS) may not be able to be calibrated. Typically, an offset can be entered into the BAS/chiller software to compensate for calibration. To do this, a calibrated sensor must be used to get an accurate temperature reading and the difference between the calibrated sensor and the BAS sensor entered as the offset. Entering an offset is not as accurate as using a calibrated sensor and replacing the sensor with one that can be calibrated is recommended.

When refrigerant pressure sensors/gauges calibration drifts, it affects the ability to diagnose problems. Inaccurate pressures can give the indications of fouling or scaling, high or low refrigerant levels, refrigerant stacking, and noncondensable gases. Misdiagnosis of these conditions can not only cost in wasted energy, but also potentially damage the chiller.

Figure 4. Dead weight tester for pressure calibration.
Pressure Calibration
There are several ways to calibrate pressure sensors/gauges including hand pump calibrators, dead weight testers (Figure 4), portable field calibrators, and laboratory calibration services. These devices and services range in price from hundreds to several thousand dollars. If a pressure calibration device is unavailable or cost prohibitive, replacement of the sensor/gauge is recommended. A typical factory-calibrated gauge can cost under $30. A high-quality, coil-type gauge that will hold calibration for a long period of time will cost slightly more.

Water Flow
One of the most common assumptions made by a facility is that water flow to the chiller is constant and always at design. Unfortunately, this may not be the case because chillers are dynamic, ever-changing models, which must adapt to the environment around them. They expand and contract from their original design and are subject to wear, tear, and age.

The impact of off-design flow can be illustrated by taking the same chiller design specifications as in the temperature example with a design kW/ton = 0.598, but with a pump that is oversized 20 percent, making the actual gpm 1,728, which would drop the ∆T to 8.33 at full load. Following the standard kW/ton equation above (using design gpm of 1,440 if flow is not actually measured), the calculated kW/ton would be 0.717, giving an apparent efficiency of 80 percent based on inaccurate data.

Figure 5. Makeshift ∆P gauge.
Measuring Water Flow
Four methods for determining flow are an inline flow meter, external flow meter, delta pressure (∆P), and ∆T. Flow meters can be high-quality turbine type, magmeter (inline), or ultrasonic (external), and give the most accurate gpm flow readings. The gpm can be determined by ∆P using a manometer or annubar. The ∆T cannot actually measure gpm, but it can determine potentially high or low flows. If an accurate ∆T at full load is greater than design, it can indicate low gpm. Conversely, if the ∆T at full load is less than design, it can indicate high gpm. If inline or ultrasonic flow meters are not available, a handheld manometer can be purchased for a few hundred dollars and is a very effective way to use ∆P to correct flow (see sidebar below).

An even more affordable but less accurate way to measure ∆P is to make your own ∆P gauge (Figure 5). Take a pressure gauge, attach a "T" pipe, connect a ball valve to each end of the T, and then connect couplings for the pressure lines to the ball valves. Connect the pressure lines to the device and the chiller's pressure inlet and outlet. Open the inlet ball valve, take a pressure reading, and then close the ball valve. Open the outlet ball valve, take a reading, and then close the valve. The difference between the inlet and outlet pressures is the ∆P.

Figure 6. Handheld manometer with couplings connected.

Sidebar: Using A Handheld Manometer To Measure ∆P

Connect the pressure hose couplings to the manometer (Figure 6). Connect the female coupling of the pressure hose that is attached to the positive side of the manometer to the male coupling on the side of the evaporator/condenser with the greatest pressure (the inlet). Connect the female coupling of the pressure hose that is attached to the negative side of the manometer to the male coupling on the side of the evaporator/condenser with the lowest pressure (the outlet). If the positive and negative are reversed, the manometer will read in negative pressure. Turn pressure valves on for the inlet and outlet and read manometer ∆P. If the manometer reading fluctuates rapidly, reduce the pressure valve flow. This will slow the fluctuation of readings. Make minor water flow valve adjustments on the evaporator/condenser until the chiller's design ∆P is met.

Important: The positive and negative (inlet and outlet) pressure connections on the chiller should be level, or the same height, in order to read the most accurate pressure. If a pressure reading is taken higher up on a pipe than its counterpart, the reading may be inaccurate.

Don Clark is with Efficiency Technologies Inc., a Tulsa, Okla.-based company that develops energy efficiency programs for commercial/industrial HVAC systems. He has over 24 years of experience in the fluid dynamics, water treatment, and chiller industries. He can be reached at 866-333-8321 toll free. For more information, visit www.efftec.com.

Publication date: 12/26/2005