Optimizing any cooling plant for minimal energy consumption is a demanding science. In many cases, minimizing chiller plant energy consumption requires modifications to the plant design, including refinement of control algorithms to assure optimal plant performance.
In this article, we will show how further energy savings can be obtained from an efficient variable primary flow chilled water (VPFCHW) system by actively compensating for system changes when operating at capacities less than design load. The discussion will be centered on countering the effects of VPFCHW central plant imbalance, single-setpoint chilled water pump differential pressure control, and condenser water pump operating pressure and flow.
VPFCHW systems achieve great efficiencies through matching chiller capacity with the cooling load. Since the cooling load is coupled with the chilled water distribution flow rate, in turn so is the chiller loading capacity. Maybe due to habits of the past or simplification of system setup, these systems are often commissioned and evaluated at the full load design and steady state conditions.
As efficient as VPFCHW systems are when operating, system design and variations in cooling loads around the distribution system can have negative effects. As the size of a plant and amount of equipment increases, it is very difficult to design out the negative system effects for both steady state and transient flow conditions. With today’s building automation system processing power and standard control devices, control techniques can be employed to automatically detect and compensate when a system starts to move outside the window of efficient operation.
Central Plant Active Balancing
A major issue that can be experienced with a VPFCHW system is the proper balance of flow through the plant and chillers. The basic premise of this design is that the chillers and the distribution are in a series loop, and flow changes are experienced by the central plant and its equipment.
As the central plant equipment experiences changing water flow due to the load demands taking place in the distribution, the variations in flow causes changes in the distribution and equipment system resistance leading to balancing issues when multiple chillers are in operation. In addition, the layout of the chillers, pumps, associated piping, and the use of different size chillers and/or manufacturer of chillers also contribute to unbalanced flow through the equipment. As chilled water flow changes and different combinations of pumps and multiple chillers are selected, the point of balanced flow shifts resulting in equipment not being loaded uniformly and operational inefficiencies. Through “active balancing,” the equipment can be maintained in balance through all load changes and when chillers and pumps are being added and removed from operation.
The chiller plant configuration shown in Figure 1 consist of two 1,800-ton 17EX Carrier centrifugal steam turbine units, two 1,800-ton York centrifugal YK single compressor units, and one 1,800-ton York centrifugal YD dual compressor unit. The existing Carrier units are mid-1990s vintage, with the York machines being new. The pressure drop in the chiller barrels is greatest with the YD machine at 25.9 ft H20, the Carrier units at 14.3 ft H20, and each YK at 5.9 ft H20, with all units having a design flow of 2,880 GPM.
The summer building load for this plant requires two chillers to operate with the possibility of a third chiller for only the hottest days of the season. There are many different combinations of chillers and chilled water pumps that the plant operators can choose to run; these various combinations cause the parallel system resistance of the chiller circuits to experience moderate to large flow imbalance.
In addition, the location of the system chilled water return lines relative to the operating chilled water pumps, the new chillers “T” layout configuration shown in Figure 2, and the piping to the chillers aggravate the imbalance condition further. Because of the many possibilities of chiller and pump combinations available, there is no way to implement a static balance which could be effective under all conditions. Adding any additional permanent balancing resistance that is not effective in all central plant conditions only results in pumping energy penalties.
As the plant goes from one- to two-chiller operation — and depending on which chillers and pumps are selected to operate — chilled water flow between the chillers can vary from 10% to as much as 45%. If a large enough flow imbalance exists, the chiller with the greater flow takes on more of the load as the other chiller lags behind in sharing the load. As flow increases further due to increased load, the chiller experiencing the greatest flow will exceed its maximum capacity first.
The cooling capacity of the other operating chiller becomes stranded and its contribution to the load is disproportionate leading to the plant not being able to maintain the leaving chilled water setpoint to the distribution, even though enough capacity is online and available. This can lead to a snowballing effect as the water temperature leaving the plant cannot satisfy the loads at the terminal equipment, causing controls valves to open further and resulting in flow increasing through the chillers. This condition may not be interpreted correctly by the operators, and they may react by adding another chiller. The large inefficiencies and operational problems introduced by the unbalanced loading of the chillers does warrant a solution. By using components typically installed and a BMS, active balancing can be implemented.
Using a BMS, analog-controlled chilled water isolation valves at each chiller, analog differential pressure instrumentation at each chiller, and chilled water plant flow meters, a control sequence can be developed to balance the plant. With the plant in operation and one chiller operating, when the operator requests a second chiller be put online, the following happens: the operating chiller unloads (i.e., chilled water isolation valve is open 100%), the chilled water isolation valve for the oncoming chiller opens to 100%, the BMS calculates the total flow for the plant and divides it by two for two-chiller operation, and the differential pressure (DP) reading at each chiller is converted to GPM and compared against the divided chiller plant flow.
The chiller with the least amount of flow has its chilled water isolation valve kept open 100%. The chiller with the greatest flow modulates its chilled water isolation valve to maintain a calculated setpoint assigned to it using its own chilled water differential pressure signal as feedback which results in balanced flow through the equipment. As the flow changes, and if different equipment is selected causing the chiller with the least flow to reverse, the program will automatically compensate by changing which chiller requires the calculated setpoint and chilled water isolation valve control. There is no reason or advantage for using fast-reacting control loops in this application, so the control process is designed to be very slow. In addition, a flow differential of 100 GPM is used with the controlled chiller GPM setpoint to prevent the system from chasing small flow changes, since a high level of precision is also not necessary. A similar sequence as described above can be used for operating three or more chillers.
Chilled Water Pump Dynamic Differential Setpoint Control
The chilled water pump differential pressure control is often set to what is required to meet the furthest load pressure drop to satisfy design requirements. Programming this setpoint to a single setting may be the simplest and quickest approach, but there are benefits that can be obtained by following where the actual load is and by adjusting the pump differential control setpoint accordingly. In addition, this strategy also compensates for central plant pressure drop changes caused by different equipment selection. This will result in reduced energy usage of the pump and chiller by flowing the amount of water necessary at the lowest pressure possible to satisfy the load under greatest demand which may not be the furthest or greatest pressure drop (i.e., design load) at the time.
In Figure 3 there are five differential sensors located in the branches to the major loads (i.e., building cooling loads). The required design DP settings from water balancing were found to be the following: Bldg. 1 – 30 PSI, Bldg. 2 – 33 PSI, Bldg. 3 – 31 PSI, Bldg. 4 – 32 PSI, and Bldg. 5 – 34 PSI. By trial and error, it was found that the buildings could be maintained at 7 PSI less than the design DP and still satisfy required partial load conditions.
One approach is to program an algorithm to evaluate all five zones. In this example, (Actual Zone DP PSI – Zone Design DP) is subtracted from the maximum pump DP setpoint in the chiller plant, in this case 40 PSI. As the zones DPs increase due to flow reduction from loads minimizing, the calculations generate setpoints below 40 PSI. The algorithm then selects the minimum calculated DP from the five zones and then assigns this value as the chilled water pump control setpoint.
It is important to mention that setpoint boundaries must be programmed to limit the variation in the control setpoint. In this case, the algorithm will not allow the control setpoint to exceed 40 PSI or go lower than 33 PSI (i.e., 40 PSI – 7 PSI). If for any reason a zone has a large increase in load or is unable to maintain its load at the current reduced DP setpoint and requires design DP, the algorithm will override the minimum selected DP setpoint and insert the maximum design setpoint as the pump control setpoint until that zone is satisfied. When this zone is satisfied, the minimum DP is reselected as the pump control setpoint again. It should be noted that fast changes to the pump control setpoint is not desirable and should be avoided.
The program used for control is required to have adequate timing, delays, and filters not to use measurements during transient periods of sudden changing conditions and to prevent erratic changes to the pump control setpoint. In a system that operates 24/7 or with systems that have diversity in load variation, energy savings will be seen during peak, non-peak hours (i.e., nights, weekends, and holidays) and as well as when loads are taken on and off line.
Condenser Water Flow Control Through Flow Coupling
Condenser flow through a chiller is usually maintained at a constant flow rate regardless of the chilled water load in a VPFCHW system. By coupling the condenser water flow requirements with the chilled water side of the system, a feed-forward open loop control can be used for the condenser water flow modulation to accomplish flow reduction. Through increasing and decreasing the condenser water flow in proportion to the chilled water flow, energy consumption and maintenance of the heat rejection system can be reduced. It can also be an effective tool in reducing the amount of chemical water treatment and evaporation/windage losses during low-load periods and winter operation.
As an example, the York YK 1,800-ton chiller requires 3,600 gpm, which is 25% higher than the chilled water flow rate of 2,880 gpm. Maintaining the 25% flow difference and using the analog DP sensors at both the chiller and condenser water bundles, condenser water analog isolation valves, and the BMS as the chilled water flow in the chilled water bundle varies downward from design load, the condenser flow can be reduced by the basic equation (Actual Chilled Water Flow *1.25) = Condenser Water Flow. The analog-controlled condenser water isolation valve will modulate to follow a calculated setpoint. This method of control automatically introduces self-balancing on the condenser water circuit regardless of how many chillers are in operation.
It’s worth noting that some caution must be taken, as there are practical limits and boundaries that need to be implemented on the range of control due to chiller minimum flow switch settings, possible increase in tube fouling, and valve speed and responsiveness to sudden increases in chiller loading. However, due to varying system dynamics, it is best to do trial-and-error testing to find the appropriate low flow boundaries.
Just as with the chiller barrels, the condenser barrels also have different operating pressure losses at design flow. The steam turbine chillers in the example presented have the largest pressure drop on the condenser water side because of the condenser water barrel and steam condenser being in series. It is not practical to set the condenser water pump DP control setpoint to a single setpoint, which in this case would be the setpoint required to satisfy the turbine machines when they may not even be selected to operate.
Since the other remaining chiller condenser DPs are also different, and to avoid operator manual adjustments that can lead to errors, a dynamic setpoint adjustment solves the problem. The condenser water setpoint is dynamically reset to maintain the lowest system DP setpoint possible by tracking the chiller in operation that has the highest DP required and adjusting the DP setpoint to match that chiller’s design differential pressure. This method can be taken further by tracking the condenser water modulating valve positions at the chillers and other equipment on the loop. An algorithm can be used to adjust the condenser water DP setpoint as valve positions continue to close below a certain percentage. This method is effective during seasons where the same condenser water loop is used for simultaneous free cooling and mechanical cooling, and in periods of reduced loads.
It is possible to successfully obtain additional energy savings from a variable primary flow chilled water plant. This involves considering not only the chiller(s) but its associated equipment. VPFCHW plant designers must first be aware that using generalized sequences will not allow for initial plant optimal operation without considering the effects of inherent VPFCHW water flow imbalance operation due to the following: operating/staging numerous chillers, their actual physical piping arrangement, their capacities (sizes), their efficiencies, loading/unloading limits, operating/staging numerous chilled water pumps/condenser water pumps, pump efficiencies, and tower(s) efficiencies, to name a few.
However, implementing plant optimization control techniques that dynamically and actively compensate for both flow and differential pressure variations during loading and unloading and during staging of chillers and pumps will help to reduce unstable control during plant load variations.
VPFCHW plant managers must seek to understand their unique plant inherent limitations, beginning with their annual load profile, prior to defining specific cost-saving plant optimization goals. Each optimization goal must be field tested to verify efficient operations at all load levels. Take advantage of the best control equipment and operating strategies available that will reduce both energy consumption and facility operating costs. Finally, VPFCHW plant personnel must be trained in all newly added optimization control algorithms and operating modes to effectively maintain daily VPFCHW plant efficiency.