For those of you that measure your experience in decades, rather than years, it is not too hard to recall a time when controls consisted of a series of electro-mechanical devices, sequenced by time clocks. The 1980's brought the centralized energy management systems, often overlaid on packaged controllers or larger pneumatic-powered systems. These early designs offered a method of gathering operating information at a central location, and operators began to modify control sequences using that direct feedback. The amount of information was limited, based on the number of expensive, field installed, hardwired points that the system had. It wasn't until the early 1990's that the cost of microprocessors dropped enough to allow their use on individual components in the system. The state of technology today provides for operating information to be accessible over LAN -based control systems.

The building blocks of today's distributed control systems are the application specific controllers. These low cost microprocessors are preprogrammed for typical sequences on terminal units like VAV boxes and fan coil units, or on packaged equipment like rooftop units or through-the-wall units. Operating information is exchanged with a central control unit, which acts as a portal to the LAN. Multiple controllers may exchange information with a single central control unit on a dedicated communication line.

General controllers are also available to control other components that lack packaged controls, or that have micro-processor controls which are not compatible with the existing control system. These general controllers are programmable to allow a customized sequence of operation.

The central control units form the backbone of the LAN, which provides access to operators through a Graphical User Interface. These are typically PC's, with network cards, modems, and headend software to allow scheduling, setpoint adjustments, and trending.

The application specific controllers can provide extensive information on the operating characteristics of the components. For example, a chiller may provide over 100 points of information, while a rooftop unit offers over 60. Even terminal units can provide substantial amounts of data, with VAV boxes and fan coil units supplying over 15 points of real time data. The general controllers provide information on 20-40 hardwired points, which can also be configured to gather critical operating data not available through the preprogrammed controllers.

In it's entirety, the amount of information which is available through these networked control systems is often overwhelming to the system operators. Most often, the trending functions are used for troubleshooting maintenance problems. This ignores the significant opportunities that are available to improve the overall operation of the HVAC system through a well-structured, long term, trending configuration.

As an example of the opportunities that might be uncovered through trending, we will review the results of a monitoring project that was recently undertaken at a manufacturing facility in New Jersey. The scope of work included a review of the chiller plant operation, and the development of recommendations to reduce plant operating costs.

The chiller plant includes seven chillers and related components, which are split into plants "A" and "B". Each chiller has a dedicated condenser water pump and cooling tower with multiple fans. The pumps and fans are equipped with variable speed drives, installed as part of a previous energy saving project, along with flat plate heat exchangers to provide "free cooling" during cold winter months. The pumps, fans, and heat exchangers are controlled by programmable general controllers, and provide feedback on motor speed, motor kw, and entering and leaving water temperatures. This same information is also available from each chiller through it's integral control panel.

Our first step is to identify the critical operating points to trend. These include not only capacity information (tons) and energy usage (kw), but also information affecting component efficiency (entering and leaving water temps, compressor differential pressure, and VFD speed). Trends are established to collect operating data on an hourly basis. The system saves a maximum of 200 points of data before rolling off, which means that the data needs to be downloaded weekly. Headend software on a laptop computer allows access to trend information through a modem. The final step is to develop a master database, which includes 80 points of data collected 8760 times, resulting in a total database with over 700, 000 entries!

Management reports are generated on a monthly basis, and include graphical depictions of critical operating information. The graph in FIGURE 1 shows the variation in total plant kw over a given month. It indicates a peak of 800 kw, with a low of just under 300 kw. Another useful graph is the load profile (FIGURE 2), which indicates the total tons delivered over time. The plant delivered a maximum of 1300 tons during the month, with a minimum of just over 600 tons.

A comparison of these two factors together yields a graph of the overall plant part-load efficiency. As can be seen from FIGURE 3, the plant generally becomes more efficient as the load increases. The bottom portion of the data spread represents the most efficient operating conditions for any given load. The goal in optimizing the operation of the plant is to minimize the spread of operating efficiencies.

One of the most difficult aspects of plant operation to model is the actual sequencing strategies that are implemented. FIGURE 4 is a graph of Building Load vs Available Tons. The horizontal lines are compiled from actual operating data, showing the sequencing strategy currently utilized. This information can be used to determine the effectiveness of the current method of sequencing.

There is also a wealth of operating information that is available from the individual system components. FIGURE 5 shows the profile of energy used by one of the chillers. It is a good indicator of chiller operating hours for the month. Another graph that can provide useful information is the part load profile. FIGURE 6 shows a part load profile for one of the chillers. As you can see, the chiller spends much of it's operating time between 40% - 70% of full load. A graphical comparison of these two factors provides useful insight into the impact of part load operation on chiller efficiency. As can be seen in FIGURE 7, chiller efficiency is dramatically reduced as it moves towards lower load operation.

An important benefit of monitoring is the establishment of a benchmark for plant operations. Operators can use this information to measure the impact of operational changes and system retrofits on the plant operating characteristics. One of the stated goals of the monitoring project is to identify O " M changes and capital projects to reduce energy costs. These recommendations are evaluated using life cycle costing techniques, to provide support for project funding. The accuracy of energy savings calculations are greatly enhanced with models developed from benchmark information. Installed costs, change in maintenance costs, and system service life are also considered in the economic evaluation.

A baseline of the plant operations was developed after completing the chiller plant monitoring for a full year. Through this analysis, a number of energy saving opportunities were identified. The most significant opportunity relates to the current method of chiller sequencing. During the interview process with the plant operators, it was revealed that the automatic sequencing software for the chillers had been disconnected, and reverted to manual control.

FIGURE 8 is a graph of Building Load vs. Available Tons. The stepped line represents the ideal sequencing strategy, based on chiller size and efficiency. The horizontal lines are compiled from actual operating data showing the sequencing strategy currently utilized. Any portions of the horizontal lines to the left of the stepped lines represents excess capacity and reduced operating efficiency. By automating the chiller sequencing strategy to minimize energy use, operating costs can be reduced by $9, 500 per year.

Plotting the tons vs WB temperature (FIGURE 9) for the two flat plate heat exchangers yields some interesting results. Heat exchanger "B" reaches design capacity of 400 tons at 20 deg F WB while heat exchanger "A" peaked at a much lower capacity. An investigation revealed that the condenser flow to HX A is significantly lower than the flow to HX B. During the design phase, a decision was made to use the existing condenser water pumps from Chillers 1 " 3. The piping run required to reach HX A is much further than HX B, thus increasing pumping head and reducing flow. By installing a booster pump to increase flow to design levels, operating costs could be reduced by $8, 000.

FIGURE 10 represents the operation of the cooling tower fans and condenser water pumps during free cooling. In the original sequence of operation, the cooling tower fans and condenser water pumps serving the heat exchangers were controlled to operate at full speed whenever the free-cooling cycle was energized. By modifying the sequence to modulate the pumps and fans to maintain a constant temperature difference across the heat exchangers, pump and fan energy costs can be reduced by $6, 000 per year.

Another part of the original sequencing strategy allowed the free cooling to be energized between November 1st and April 1st. Based on information derived from the previous graph of heat exchanger capacity vs. wet bulb temperature, there are additional hours during the year that the free cooling cycle would be beneficial (FIGURE 11). By extending the free cooling cycle from October 1st to May 1st, an additional $2, 500 could be saved.

Implementing these changes will result in an annual operating cost reduction of $26, 000 per year. The annual cost to maintain the monitoring contract is $3000 per year. With an estimated installed cost of $25, 000 and a ten-year service life, the project has a net value of $ 75, 000. These opportunities would not have been uncovered without the information provided through the monitoring program. It is a low cost method to gather data with which to continue the process of optimization.

FIGURE 1: PLANT KW

FIGURE 2: PLANT TONS

FIGURE 3: PLANT PART LOAD EFFICIENCY

FIGURE 4: CHILLER PLANT SEQUENCING

FIGURE 5: CHILLER ENERGY USE

FIGURE 6: CHILLER PART LOAD PROFILE

FIGURE 7: CHILLER PART LOAD EFFICIENCY

FIGURE 8: BUILDING LOAD VS AVAILABLE TONS

FIGURE 9: HEAT EXCHANGER OPERATION

FIGURE 10: FREE COOLING PUMPS AND TOWER FANS

FIGURE 11: OPERATING HOURS BELOW 50 Deg F WB