The latest improvement involved the upgrading of the chiller systems at the company’s 100-acre research and development campus.
Two central chiller plants provide cooling water to buildings across campus. Plant #1 consists of five 1,050-ton, three 800-ton, and one 500-ton constant-speed electric centrifugal chillers, plus one 950-ton steam-fired double-effect absorption chiller.
Plant #2 consists of four 1,050-ton and one 500-ton constant-speed electric centrifugal chillers, plus one 950-ton steam-fired double-effect absorption chiller and two 680-ton steam-fired single-effect absorption chillers.
Chilled water had been distributed from each plant via a primary/secondary/tertiary pumping system. Dozens of pumps circulated water in the primary loop within the plant (constant flow), secondary loop from the plant to the buildings (variable flow), and tertiary building loops (variable flow).
Mehta’s team determined that the pumping system was over-designed considering the technology available today. The result was excessive pump horsepower and wasted energy throughout the cooling season. He made a presentation to management that proposed converting the chilled water systems to a variable-volume primary-only pump system, by reconfiguring piping in the plant and at the buildings and upgrading plant controls. This would eliminate the need for 42 pumps, simplify operation of the system, and save pump energy as well as substantial chiller energy with improved staging.
Within each plant, one constant-speed pump (ranging from 40-75 hp) would run per operating chiller, circulating water through the primary loop. The secondary pumps’ variable-frequency drives (VFDs) modulate their speed (flow) to maintain a differential pressure set point in the system. Most of the buildings have variable-speed tertiary pumps, which pull water from the secondary loop supply and send it through the air handling units (AHUs) in their buildings. After the job was completed, only one set of variable-speed pumps was needed to provide flow for cooling generation and distribution to the buildings.
“We estimated the total cost of the project at $1.8 million and an annual savings in power usage of approximately $400,000,” said Mehta. “That would make a return on investment of more than four years. But, we also found a way to make it more affordable — New Jersey’s Clean Energy Program (CEP).”
The CEP offered a program called Large Energy Users Pilot that offered financial incentives for entities to improve their energy efficiency, and the pharmaceutical manufacturer certainly qualified as a large user. Mehta and his team worked up energy-efficiency plans detailing the pump replacement, submitted them to the state, and they were accepted. They qualified for more than $700,000 in incentives.
“That incentive dropped my client’s investment to just over $1 million and shortened the ROI to about two-and-a-half years,” said Mehta. “The project was approved and began about a year ago. It was completed in May.”
In addition to the electrical energy saved by eliminating 42 constant flow pumps, the variable-flow pumps further reduce water flow and energy usage throughout the newly revised system, because only the water flow required by the load is chilled and delivered. With a variable-flow primary-pumping system at the majority of loading conditions, no chilled water is returned back to the chiller, without being used by the load.
Varying the water flow through the chiller has many benefits. It allows production-pump energy savings and compressor energy savings during partial load operation. The quick unloading capabilities of variable-capacity compressors makes varying water flow through the chiller possible, because a consistent outgoing water temperature can be maintained over a variety of loading conditions. This reduces the likelihood of nuisance trips of the chiller safeties, which prevent the chiller from freezing, and it also optimizes performance of the chiller system.
It is necessary to include a supply-water bypass with modulating valve and return water-flow sensor with variable-flow primary pumping systems. During low load conditions this system ensures a minimum water flow through the chiller. The modulating valve and return water-flow sensor only allow the bypass to be used at low-load conditions to maximize energy efficiency at other loading conditions.
Varying the water flow through the chiller also minimizes the efficiency losses during low ΔT syndrome conditions. To combat a low ΔT, a variable-flow primary-pumping system can vary the water flow through the chiller to optimize efficiency.
Increasing chiller efficiency is certainly important, but equally important is the ability to measure the flow of the water throughout the system.
“You cannot manage what you don’t measure,” said Mehta, “so it was critical to have accurate flow measurements. But we needed to get it at the most economical way possible without sacrificing accuracy. In a new plant, I typically recommend magnetic flow meters because they offer good accuracy. But this was an old plant, and mag meters are intrusive, so the plant would have to be shut down and the pipes would have to be cut to install them. That would be expensive.”
Tim Kusters, account manager, Technical Devices Inc., a process control instrumentation representative, had been supplying ultrasonic flow meters, among other instrumentation, to the pharmaceutical company’s facility manager for years. That is where he heard about the WMGroup chilled-water optimization project.
“They had originally specified mag meters,” said Kusters. “The Flexim flow meters we recommended clamp directly on to the pipes with no need to shut down the plants and cut into the pipes. Also, there were some flow measurement points with very short, straight runs. Clamp-on ultrasonic meters are capable of using multiple beams which mitigate the effects of short pipe runs. With this capability, we were able to guarantee flow meter accuracy comparable to, or better than, mag meters. Mehta and his team chose the ultrasonic meters because they provided a significant savings in installation costs and provided excellent accuracy. While we knew that this application was for the chilled-water optimization, we were told they would likely expand the project in the near future to include the monitoring of condenser water from the cooling tower. So we installed dual-channel electronics everywhere to take advantage of the fact that we could later tie in the condenser water on the second channel of the electronics/transmitter. The 25 Flexim meters we installed communicate with the plant’s SCADA system via BACnet. When the condenser water was added, it was much more economical because they didn’t have to wire and power an additional transmitter because a two-channel ultrasonic meter can measure the flow in two pipes.”
The technique that Flexim’s ultrasonic flow meters use is called transit-time difference. It exploits the fact that the transmission speed of an ultrasonic signal depends on the flow velocity of the carrier medium, kind of like a swimmer swimming against the current. The signal moves slower against the flow than with it.
When taking a measurement, the meter sends ultrasonic pulses through the medium, one in the flow direction and one against it. The transducers alternate as emitters and receivers. The transit time of the signal going with the flow is shorter than the one going against. The meter measures transit-time difference and determines the average flow velocity of the medium. Since ultrasonic signals propagate in solids, the meter can be mounted directly on the pipe and measure flow noninvasively, eliminating any need to cut the pipe.
When first introduced, ultrasonic meters were met with skepticism, as one problem justified that doubt. The couplant grease that sealed the transducer to the pipe would migrate out over a few years and the meter would fail. Flexim solved the problem by developing a non-grease solid-pad couplant. Because of their success in the field, ultrasonic flow meters are now accepted as a highly accurate, nonintrusive measurement system.
To verify the electric energy savings resulting from the variable pumps, a whole-facility approach used electric utility data and cooling load trend data. Energy use (kWh) was normalized for the cooling generated in the same period (ton-hours). The efficiency of the cooling system before and after the improvements (kW/ton) was compared.
A cooling-load profile that was developed for the facility based on available trend data from both chiller plants, which spanned from 2010-2011, was the basis of the energy-savings calculations. The total ton-hours of cooling used in each month was plotted against the corresponding monthly energy use from the utility meters. A linear regression showed the base monthly energy use that is not attributed to cooling. This energy was subtracted from each month and the remainder was totaled for the whole year, yielding the annual electric energy consumption of the cooling system. This was divided by the total ton-hours generated during the same period to get the average system efficiency in kW/ton.
After the Project
After the variable pumps were in use, the energy management system trended the cooling load on the chiller plants for a minimum of six months. The monthly ton-hours were totaled and plotted against the coincident electric energy use from utility meters. Another regression was run and the base monthly energy use was subtracted from the utility totals. The remaining electric energy use was totaled for the period, along with the ton-hours, and the subsequent variable-pump cooling system efficiency was calculated.
“The savings were in line with what we had estimated when we came up with estimated savings of $400,000 per year,” said Mehta. “One example of increased efficiency can be seen in chiller usage during the unusually hot summer of 2012. In the past, our client would need all of his chiller capacity in the summer, and have none in reserve, hoping no system shut down. Last summer, with the upgraded chiller systems in use, they had at least three systems in reserve.”
Publication date: 8/5/2013