In the early days of refrigeration, companies used massively oversized heat exchangers with large-diameter tubes, and chillers often were placed in series. This configuration was economically viable, and competitive pressures were comparatively low.
Parallel refrigeration emerged as a viable alternative for several reasons. First, with easily configured rolled-groove couplings, it was easy to add capacity by adding refrigeration system “modules” to common headers in applications where all of the refrigeration plants were in parallel to the main system headers. Then, the process or cooling load side began to change how chilled water was used. One example was the variable air volume (VAV) designs, which could use higher chilled water temperatures and remain effective across a range of load conditions. While it is now common to see technology such as direct digital control (DDC) that compensates for the chilled water usage based on real-time process requirements, this level of control was not emphasized or even possible in earlier times. Chilled water systems were designed to ease the work output of the chiller with generous approach temperatures and minimal temperature changes.
Today, chilled water production, delivery and usage systems are evaluated together, along with flow rates, pipe sizes, pressure drops, compressor type, end-point coils, heat exchangers and overall system “lift.” These ingredients, as well as variable flow schemes, are changing the way engineers design systems to optimize initial cost, operational efficiency and owner complexity. New control components, a steeper heat transfer to fluid velocity curve, wider acceptable flow ranges across heat transfer surfaces, and the implied heat exchanger and control requirements at the chiller are forcing engineers to rethink parallel vs. series refrigeration once again. For many applications, equipment using true counterflow series arrangements can provide significant short- and long-term advantages.
Maximizing Infrastructure, Minimizing InefficienciesIn series-series arrangements, both the evaporator and condenser circuits are in series and in countercurrent flow, so that the downstream chiller producing the coldest chilled water temperature rejects heat to the coldest condenser water temperature. Lift is an important consideration. On the refrigerant side of the equation, lift refers to the difference between the evaporator and condenser refrigerant pressures. Lift is approximated by the temperature difference between the leaving evaporator and the leaving condenser water temperatures. The higher the lift, the more energy is required by the compressor. A lower lift can be achieved by creating warmer temperatures in the upstream evaporator and equalizing lift across the compressors by using counterflow. Lift does not depend on the evaporator's temperature differential (∆T).
In series configurations, each compressor performs only a part of the total system requirement and uses temperatures with the largest possible approach, automatically reducing lift. Because series-series configurations reduce lift, they also reduce compressor energy use. Figure 1 shows a simple two chiller series counterflow arrangement and the associated reduction in effective compressor lift. Screws, scrolls and reciprocating compressors are positive-displacement systems, and their efficiency is sensitive to lift. Be careful not to confuse chilled water ∆T with lift; a system can have decreased lift but still have a “normal” ∆T for each unit, and thus a wider overall ∆T for the set.
The system as a whole must be considered because pumping energy can be higher than the savings realized at the compressor in systems with high heat exchanger pressure drops. In series arrangements, each heat exchanger sees all of the flow for the entire system. This situation is both bad and good. The good news is that modern heat exchangers use enhanced surfaces that have a wider total range of maximum to minimum acceptable flow, so the available turndown is increased. The bad news is that because all of the flow enters every heat exchanger, the pumping system must accommodate the increased pressure drop when operating at full load. Shell-and-tube heat exchangers often are selected because the pressure-drop-to-velocity relationship is less favorable in brazed plate designs, resulting in a higher pressure drop at the same fluid velocity. Pressure drops can be further reduced using single-pass rather than multiple-pass shell-and-tube heat exchangers.
Another consideration is that because pump energy is approximately proportional to the cube of the flow, variable flow schemes reduce the increased pressure drop of a series configuration. Pump horsepower is critical because enough compressor benefit must be obtained to offset the extra pumping power. Lower flow rates reduce the pressure drop and pump power. Lowering the setpoint to operate at lower temperatures can also be beneficial -- a 3oF (1.6oC) reduction in the setpoint can lower the pumping power requirement by 50 percent or more and still meet heat load demands. Variable flow also minimizes the increased head penalty at most load conditions.
In parallel refrigeration, some significant issues come into play. In a constant flow scheme, water flows through each heat exchanger at roughly the same temperature. Therefore, chillers in parallel must produce the coldest water required for the system while rejecting the heat to the warmest condenser water. This design ensures higher energy costs because each compressor must do the maximum amount of “lifting” (figure 2).
In practice, some variation in flow rate occurs as some of the water passes through the first parallel chiller, reducing flow to subsequent parallel chillers. However, because the coefficient of flow (cV) authority belongs to the header in properly designed systems, this flow difference is minimized.
Constant flow systems in parallel configurations suffer from a more significant issue. When operating at part load with a parallel chiller turned off, the warm return water simply passes to the chilled water supply side without being cooled. This results in mixing and raises the overall system chilled water temperature being delivered to the process. The solution is to lower the setpoint of each operating chiller to compensate for the mixing effect, thereby increasing the lift and the energy required to meet the part load demand. These two effects, compressor lift and mixing, combine to make parallel refrigeration costly to operate during part load in constant flow systems. The conclusion is that parallel systems must use variable flow methods to avoid creating circumstances that drive up energy costs.
When a parallel chiller is not required, the flow should be reduced as well. Simply shutting off the water flowing through any non-operating parallel chiller forces all of the water to pass through the operating modules, thereby vastly increasing the fluid velocity, pressure drop and energy required to pump the water. Remember the cube formula? This effect is somewhat mitigated when a system has two refrigerant circuits and one water circuit because some cooling takes place, and controls can stage compressors until they reach a point where whole modules must be staged off. However, this solution suggests that using more modules with more circuits is better from an energy efficiency standpoint, and the initial equipment costs go up as a result. These limitations are avoided in many series-series configurations.
What to Turn Off FirstA caveat exists for parallel variable flow methods when a refrigeration system must operate across a wide load range or when refrigeration demands change quickly. In constant flow systems, it is more efficient to operate one of the two refrigerant circuits in each chiller module to minimize the chilled water mixing problem described previously. In variable flow systems, the proper control method is to stage off entire chillers as quickly as possible and shut off the flow across their evaporators to eliminate the mixing problem. This action is easily accomplished by using common automatic isolation valves at each unit. However, this design is not possible in series systems because any isolation at the module prevents flow from reaching downstream units. Therefore, with series systems, three-way bypass valves are used instead of two-way isolation valves.
With a traditional single evaporator system, the minimum acceptable flow rate is that across the evaporator, perhaps to as low as 40 percent of the total system flow. (This flow rate is velocity driven. It is common to select heat exchangers to operate between 3 to 11 ft/sec, but in some cases, fluid velocities as low as 1.5 ft/sec may be adequate to produce turbulent flow.) But in a parallel system with variable flow, the staging method prioritizes shutting off whole chillers, not individual circuits. Therefore, with isolation or bypass, the lowest possible flow rate is the flow rate across one single parallel unit, not the whole system. If the lowest acceptable flow rate across a heat exchanger is 40 percent, and there are four parallel units, the total flow range across a single evaporator is 100 percent down to 40 percent. In a properly configured parallel variable flow system, the total range is from 100 percent to 12.5 percent. According to the cube formula, the resulting energy use is significantly lower than this simple numerical representation indicates.
Let's Get Really EfficientWith series methods, each chiller will see different temperatures depending on where in the series arrangement the chiller sits. If the flow is counter-current, all of the heat exchangers will experience the same approach; the mean temperature difference will remain constant. If the system has four series chillers, each chiller compressor should be optimized around the prevailing temperatures for that unit. Using the same compressor in each series unit eliminates some of the energy savings because each compressor is not optimized for the refrigerant pressure and volume it will experience.
Engineers are increasingly dropping chilled water temperatures, reducing flow rates and raising the ∆T in a process to realize these efficiencies. Gone are the days in which overall system efficiency is relegated only to the chiller. Now, the chiller must accommodate a lower chilled water temperature to increase the capacity of coils and process equipment and to dissipate heat and/or control humidity. Infrastructure costs in pump horsepower and pipe size also can be reduced when a system delivers colder water at lower flow rates. Pumping cost savings can more than compensate for chiller efficiency losses in well-designed systems.
For larger systems with higher capacity requirements and longer pipe runs, it is less expensive to deliver colder water at a lower flow. Because more chiller energy is required to create colder water, engineers must ensure that increases in system efficiencies offset any increase in compressor energy requirements. For example, in a system in which the temperature is reduced from 45 to 37oF (7 to 3oC) and the condenser flow is reduced from 3 gal/min/ton to 2 gal/min/ton, the chiller heat exchangers might be designed to operate at the lowest economical approach of 3 to 5oF (1.6 to 2.8oC) instead of the traditional 10 to 12oF (5.5 to 6.6oC). It should also be noted that flooded evaporators can be used instead of direct expansion evaporators to produce economical 2 to 3oF (1.1 to 1.6oC) approach temperatures. Modern pressure-compensated control valves are available at the point of use to control chilled water flow based on real demand and are pressure independent, meaning that they self-adjust to the changing pressures of variable flow methods. Microprocessors can sense demand changes and can vary pumping power and operate control valves to maintain the proper heat exchanger velocities automatically.
Heat recovery from the condenser water is much easier to obtain in series configurations because of the higher discharge condenser water temperatures, especially at reduced condenser water flow rates. Higher condensing temperatures do increase compressor lift, but savings are offset by reduced pumping costs. In fact, if heat recovery is desired and the application has a simultaneous hot water and chilled water demand, series condenser water at reduced flow might be the most economical approach. This approach requires the chiller to have condensers optimized for lower flow and capable of creating the higher compressor lift required to deliver rated capacity without a significant efficiency loss.
In variable flow arrangements, system stability is also enhanced in series configurations because transient flow disruptions are minimized during chiller on and off staging, and the controller can more accurately and rapidly respond to changing load conditions without the risk of overshooting the system's temperature setpoint, which is more common in parallel designs. Control designs for parallel configurations often employ long-delay timers to mitigate this issue, but delay equals cost.
Another consideration is seasonal latent load changes. In humid weather, circulating lower temperature water to coils is an efficient way to ensure proper dehumidification; however, in low humidity or off-peak conditions, higher temperature water to the coils might be sufficient. Advanced controls can vary the chiller setpoint to reflect the optimum temperature at the coils.
Many of these approaches to maximizing chilled water system efficiency have their roots in larger chiller plants, where higher flows and pump power are common and costly, and can therefore provide higher benefits and shorter paybacks. Further, the benefits do not stop at just the chiller and pump cost reductions; using series configurations also provides a reduced electrical infrastructure and environmental impact. The same efficiencies can easily be realized with carefully designed smaller systems using a variety of compressor technologies, especially new magnetic drive, friction-free, variable speed centrifugal designs that can operate at unmatched part load efficiencies. As innovative manufacturers leverage new technologies and as chiller system efficiencies continue to improve, owners can expect increasing benefits from using correctly applied series and parallel configurations to minimize initial and operational costs. PCE
Jackson L. Ball is director of business and technology at ArctiChill and Freeze Co. Systems, Brampton, Ontario, Canada, a manufacturer of chillers and process cooling systems. Susanna Hanson is senior product support engineer at Trane Commercial Systems, LaCrosse, Wis., a company that provides indoor comfort systems and facility solutions for industrial, commercial and residential buildings. For more information, contact Arctichill and Freeze Co. by calling (404) 231-0720; visiting www.www.arctichill.com or www.freezeco.com; or e-mailing firstname.lastname@example.org. Contact Trane Commercial Systems by calling (608) 787-2000; visiting www.trane.com; or e-mailing email@example.com.