firm profile
contact us
dema project



Name of Project:
Department of Emergency and Military Affairs

United States Army

National Guard

CMSB, Building #L4525

Marana, AZ

Building Description:
Constructed in 1993, the facility is replete with energy conservation strategies of the time. Variable pumping and airflow schemes, variable flow capabilities of cooling tower fans, and four-pipe hot water and chilled water distribution systems should have made for an efficient HVAC system. Direct Digital Controls (DDC) have since replaced outdated controls, and should have been effective in generating even more utility cost reductions.

However, as is often the case, the facility’s HVAC system was substantially oversized and as a result required exacting controls capabilities for efficient operation, yet a lack of commissioning and improper application of controls strategies exacerbated the problems. In most cases, a “bigger is better” approach to HVAC systems is counterproductive and results in the system’s using excessive energy and creates poor space conditions. Oversized HVAC systems typically have any number of deficiencies, but the good news is that once solutions have been devised and implemented the energy costs fall dramatically and space conditions improve significantly.

Analysis of historical energy data is rather disturbing. Our HVAC analysis did not account for the 100-HP hydraulic pump set serving the flight simulators, yet even so the energy consumption is staggering if compared to like-use facilities with similar HVAC systems. Utility charges include .0615$/KwH and $15.25/Kw for demand. The demand for the facility has ranged from 534 to 403 KW over the past 3 years, a very high number, but not surprising considering how many HVAC electrical loads – fan motors, pumps, water chillers, cooling tower fans - are operating unnecessarily to condition the facility.

The building occupancy profile is essentially office with air-conditioned “warehouse” space (the simulator bays) and their supporting computer rooms. As will be discussed later, the electrical loads (and resultant HVAC loads) in the computer rooms are quite small. Operating schedules for the hydraulic pump set and the ability for continuous operations (8,760 hours per year) make estimates an inexact science. Nevertheless, based upon our experiences, considering the energy conserving devices used by the HVAC system and the loads they serve, and making assumptions for the hydraulic pump set, we estimated that the facility was needlessly expending a significant amount of its current utilities.

Systems Subject to OptimissioningSM
Generally speaking, the required work for this project focused exclusively on the chilled water aspects of the heating, ventilating and air conditioning (HVAC) system, with emphasis on the operation of the computer room a/c (CRAC) units relative to both mechanical refrigeration and waterside economizer modes of air conditioning. It was expected that the determination of staging for the CRAC units would be an iterative process to determine the most efficient method of control: staging on multiple units at higher supply air temperatures, running the minimum number of units at lower supply air temperatures, or a combination of the modes as is expected.

The variation in supply air temperatures from the CRAC units will be a function of the chilled water supply temperature. Currently, warmer than design water is continuously delivered to the units, caused by poor piping practices which are creating a reverse decoupling of the secondary chilled water. As now intended, the chilled water temperature will be warmer intentionally, either due to the use of a chilled water reset schedule for the water chillers, or operation of the plate frame heat exchanger. Warmer water will entail warmer supply air, necessitating additional air, which is to be supplied by the redundant CRAC units staged by soon to be installed controllers.
In addition to the computer rooms CRAC units, other systems optimissionedSM included:
• The entire chilled water delivery system, including the three water chillers, three primary pumps, and two secondary pumps that are equipped with variable frequency drives.
• The condenser water system including the three condenser water pumps and three cooling towers and their respective fans equipped with variable frequency drives.
• The air distribution system, particularly VAV air handling unit #3 and all associated VAV boxes.
• Air handling units #1, 2, 4 and all fan coil units will be evaluated for cooling load requirements and chilled water and air flow rates will be adjusted during the rebalance phase of the project.
• The entire DDC system will be evaluated and modified if deemed necessary

HVAC System Description:
Chilled water and condenser water system:
• Three (3) nominal 110-ton reciprocating water-cooled water chillers selected for 55-degree entering, 45-degree leaving chilled water.
• Three (3) cooling towers selected for 95-degree entering, 85-degree leaving condenser water temperatures with 340 GPM at 79-degree wet bulb (WB) temperature. Each tower fan is equipped with a variable frequency drive.
• Three (3) 5 HP chilled water pumps selected for 270 GPM at 40’ total head pressure.
• Two (2) 20 HP secondary chilled water pumps equipped with variable frequency drives, selected for 810 GPM and 50’ of head each. One pump is redundant.
• Three (3) 5 HP condenser water pumps selected for 340 GPM at 40’ total head pressure.
• One (1) plate/frame heat exchanger, approximately 73 tons capacity. Heat exchanger is selected for a 3-degree approach, with 47 degrees entering the heat exchanger and 50 degrees leaving to the chilled water system.

Air distribution system:
There are four (4) air-handling units providing space temperature control. Air handling unit #1 serves the East bay of the facility, air handling units #2 and #4 serve the West simulator bay(s). This bay (West) is only now being outfitted with simulators, and the corresponding computer room put into service – including the installation of all computing hardware - to provide the requisite computer simulations for a fully functional simulator system.

The air-handling units’ chilled water-cooling coils were selected (design day) for 45-degree entering and 55-degree leaving water temperature, with a resultant deck temperature of approximately 55-58 degrees. Air handling units #1, #2, and #4 are equipped with hot water coils, and are constant airflow. Air handling unit #3 serves the second floor office areas and is variable flow, utilizing the aforementioned VAV boxes for zone temperature control. Four-pipe fan coil units serve the remaining zones’ (essentially the “Lobby” areas) heating and cooling requirements.

Computer rooms:
The dual “computer rooms” are an interesting and divergent subset of the HVAC operational challenges at the facility. Both rooms appear to have at least one redundant computer room air conditioning unit (CRAC) installed. The East room currently uses four (4) nominal 22-ton units, and the West room utilizes six (6) nominal 15-ton units and has two (2) additional air-cooled (remote condenser) DX units, nominal 5 tons each.

Both rooms utilize and under-floor supply air delivery system, and all CRAC units are complete with chilled water and hot water coils, infra-red (IR) humidifiers, and have relatively simple temperature and humidity controls. Both the chilled water and hot water control valves are 2-way modulating.

DDC system:
The current control system is Direct Digital Control (DDC), manufactured by Alerton Controls, and is currently being maintained by Climatec, who also installed and commissioned the system. Our observations of the control system indicate that it is quite capable of providing the level of control required to maintain comfortable space conditions, adequate chilled water and hot water capacity for computer room operations, and do so efficiently.

The system is a fairly sophisticated computerized distributed processing system that uses “Visual Control Graphical Programming” language. This language is extremely versatile when an experienced programmer designs the functional programs, provided the correct sensors and inputs are in place. The resolution of the temperature/analog (input) sensor(s) appear to be more than sufficient to provide excellent temperature control and system response

Systems Deficiencies and Solutions

The system had a 43-degree chilled water setpoint, considerably below the design leaving of 45 degrees leaving, and approximately 48-degree return chilled water – an obvious indicator that the system is severely over-pumped. Moreover, it appeared as if approximately 46-degree water is actually being delivered to the chilled water coils throughout the facility, almost certainly due to “reverse” decoupling thru the primary-secondary decoupler. In other words, the return chilled water – itself extremely low at approximately 48 degrees – was mixing with the 41 degree chilled water from the chillers, affecting a delivered chilled water temperature of approximately 46 degrees.

Such reverse decoupling is common for primary-secondary piped systems that employ full-size, front decoupled methodology. It was evident that one of the two available secondary pumps is intended to be redundant, yet both were operating and both are operating at very high speeds to maintain the differential pressure setpoint. This high DP setpoint was causing other energy wasting, space temperature control issues including affecting the chilled water control valves’ ability to modulate flow effectively. The location of the differential pressure (DP) sensor was also observed to be less than ideal, and is controlling to much too high of a setpoint.

The high chilled water system pressures were causing such issues, and the resulting performance penalty was three-fold. First, sub-cooling of the air will require additional reheat to maintain zone temperature setpoint, obviously an inefficient operation. Second, some zones may not have the ability to achieve setpoint during the winter (heating) seasons, creating uncomfortable conditions and likely occupant complaints. Third, the pumps must run at higher speeds, using more energy, to maintain the high DP setpoint. Conversely, in summer (cooling) months the inability to deliver cold enough water, due to the reverse de-coupling described above, in order to dehumidify the air will result in “stuffy” space conditions. Comfort and productivity issues have been well documented to be directly proportional, so the net result becomes higher energy costs and lower productivity.

Chilled Water System Deficiencies:
The initial study of the chilled water system revealed that the entire system was grossly oversized, poorly designed and completely out of control. Our estimates indicated that approximately 120 tons of chiller capacity would be required on a design day, yet 330 tons was actually installed. As discussed above, an oversized system is typically counterproductive from an energy perspective unless a concentrated effort is made to eliminate wasteful operating practices.

Site visits revealed that the chillers had different setpoints: Chiller 1B was operating with a setpoint of 43- degrees, had 12,715 starts and 25,313 total run hours. Chiller 1A’s setpoint was 41-degrees, had 25,639 starts and 29,459 run hours. Chiller 1C also had a 41-degree setpoint, 23,630 starts and 29,452 total run hours (all above data thru 2003). The number of starts per run hour is cause for concern, as the units are obviously starting/stopping much too frequently, and the total run-hours for all (3) machines shows that 330 tons is commonly being produced to satisfy a 120 ton load. This phenomenon would be expected from a system that experiences very low chilled water system temperature differentials such as existed at this facility.

Lastly, the original plate/frame heat exchanger was selected for a 3-degree approach between entering cold (condenser water) and leaving hot (chilled water supply), limiting the number of hours that the facility could be satisfied using “free” cooling. It was also determined that the existing control sequence of operations had several flaws which would prohibit plate/frame heat exchanger operations regardless if the ambient conditions were suitable for “free” cooling operation.

Chilled Water System Solutions:
Installing a new 2-position isolation valve in the existing de-coupler, and installing a new, smaller sized de-coupler in a much better location – from the end of the primary pumps’ header to the main return line - solved the “reverse” de-coupling issues. A bi-directional flow meter was installed in the de-coupler to monitor both flow rates and direction of flow. As expected and intended, the flow thru the de-coupler is now from supply to return, depressing the return water temperature and allowing the chillers to unload.

A new 1-degree approach plate/frame heat exchanger was installed to replace the existing 3-degree heat exchanger, extending potential operating hours. During the optimissioningSM phase, the chilled water diverting valve responsible for directing the chilled water flow to the heat exchanger was found to be defective and was subsequently replaced.

The central plant sequence of controls was rewritten to correct existing deficiencies, and applicable setpoints and timeouts – i.e. the time delays going into and coming out of plate/frame and chiller operation - were modified and optimissionedSM. As the HVAC system is substantially oversized, duty cycles were established for all chillers, cooling towers, and primary and secondary pumps as shown below. A chilled water temperature reset schedule was established based the ambient wet-bulb (WB) temperature.

The heat exchanger setpoint and the chilled water reset schedule upper limit are close in proximity to the actual deliverable chilled water supply temperature. The cooling towers have approximately a 6-degree approach, and coupled with the 1-degree heat exchanger installed, the chilled water supply temperature at 52 degree WB (ambient) will result in approximately 59-degree supply chilled water. Above 52-degree WB, the chillers will start and produce approximately 57-degree supply chilled water. OptimissioningSM observations indicate that the Data Rooms, when enabled with temperature staging discussed below, can maintain temperature limits at these supply chilled water temperatures.

The chilled water system differential pressure (DP) control was also optimissionedSM. A system pressure of 12.5 PSIG was being maintained prior to optimissioningSM the HVAC system. Subsequent to improvements made in the air systems and CRAC’s as discussed below, the final system setpoint is now 3.75 PSIG.

Prior to the HVAC improvements and subsequent optimissioningSM, the chilled water system DT was extremely low (approximately 3-4 degrees) and caused additional chiller systems to be started in response. This is a common requirement of most systems, as most experience low system DT’s and the resultant inability to load the chiller to its full load capacity. For example, the chillers at CMSB are rated for approximately 113 tons at a 10-degree (55-45) chilled water DT.

The new central plant piping design and new sequence of operations resulted in the chilled water loop receiving warmer than design (45-degrees) entering water temperatures during plate-frame operation. Improving chiller plant - specifically chilled water piping/pumping/controls - operations eliminated the need for the Sears dehumidifiers, as the ability to deliver cold enough chilled water (during high temperature/humidity ambient conditions) to the CRAC units to dehumidify the computer rooms was greatly improved. At project closeout, unusually high ambient conditions existed, yet the entire facility was operating on the plate/frame heat exchanger for the majority of operating hours. Beginning on 3/15/2005, the run-hours of the water chillers were reset to zero so that chiller run times could be tracked from the completion of the project.

Condenser Water System Deficiencies:
During optimissioningSM, it became apparent that the cooling towers were seriously underperforming. Observations and readings included:
• Cooling tower fan motors operating at approximately 40% of rated full load amps even though the variable frequency drives were at 60Hz.
• One operating cooling tower could not achieve predicted wet-bulb (WB) approach results of approximately 8-degrees between outside air WB and supply condenser water temperature. Two cooling towers were required, ½ of the total condenser water flow each, to achieve this approach.
• Extremely restricted airflow thru the cooling tower fill, as would be expected considering the above.

The DDC sequences controlling the condenser water system were less than ideal and essentially precluded the use of the heat exchanger due to their programming logic and setpoints. Also, during the optimissioningSM phase, it was discovered that three modulating condenser water isolation valves were defective; one was located at cooling tower #1B, one at cooling tower #1C, and one at Chiller #1A.

Condenser Water System Solutions:
The condenser water valves were replaced, and the cooling towers were also replaced with an identical model in order to facilitate an expedient change-out and keep the central plant in operation at all times. The new cooling towers were incorporated into the new central plant sequences and at project closeout were performing extremely well.

The cooling towers can produce approximately 6-degree approach to ambient WB as described above, enabling a quite high heat exchanger changeover setpoint of 52 degrees. Programming code was added so that when heat exchanger mode is initiated, two (2) cooling towers are energized to suppress the cooling tower supply temperature more quickly than one tower. The cooling tower variable frequency drives’ PID control loops were modified to eliminate “hunting”, and the condensing water system is performing as designed and intended at this time.

The timing of critical actions was established during optimissioningSM. For example:
• A 10-minute period was established to allow the condenser water system to achieve the heat exchanger setpoint
• The 10-minute delay is observed before switching any valves needed to enable the heat exchanger mode of operation
• Two cooling towers are enabled on a call for heat exchanger, enabling the condenser water loop to be sufficiently depressed during the 10-minute delay

The new sequence of operations for the DDC system also included changes to the condenser water system. The cooling towers have a common supply water header, so any tower can serve and chiller system. As we expect two cooling tower to be required at the very most, and one cooling tower required the vast majority of the time, the cooling tower were set up to rotate Lead-Lag-Standby on a weekly basis in an attempt to maintain a consistent water treatment throughout.

The condenser water pumps are dedicated to their respective chillers and will rotate in conjunction with the chiller duty cycles. Chiller #1A is to be permanently dedicated as Emergency (Standby), yet its condenser water pump (#3A) is dedicated to the heat exchanger and as such is enabled on a call for heat exchanger operation.

Computer Rooms Systems Deficiencies:
Similar to the chilled water system, the Computer Rooms Systems are also greatly oversized. All of the four nominal 22-ton units (88 total tons) were in operation in the East Computer Room during all site visits and prior to any HVAC system improvements or optimissioningSM, indicating a lack of control capability and a ready source for energy conservation. Also, a number of residential style Sears brand dehumidifiers were located throughout the East room, with personnel indicating that they are required during the monsoon (high humidity) season. We believe this to be caused by the “reverse” decoupling discussed earlier, as the entering chilled water temperature to the CRAC units’ chilled water coils is warmer than design, and not sufficiently cold to dehumidify the air.

Documentation from the original severs (East room) shows a total heat of rejection of 487,100 BtuH, or approximately 40 tons. This is assumed to be the maximum possible heat rejected by all servers, but the service entrance section (main power) dedicated to the servers showed many unused circuits. Field measurements indicate an approximate heat load of 170,750 BtuH, or 14.2 tons in the East Computer Room, and 102,450 BtuH or 8.5 tons in the West Computer Room currently under construction. Additional cooling loads will include lighting, people and infiltration loads, but we expect that these are minimal. Field measurements of the service entrances providing power to the server racks is as follows:
A B C Average
E Room: Service 1: Current 95 102 113 103 Amps
Voltage 198 198 198 198 Volts
Service 2: Current 22 28 19 23 Amps
Voltage 198 198 198 198 Volts

W Room: Service 1: Current 88 78 62 76 Amps
Voltage 204 204 204 204 Volts
Service 2: Current 10 6 2.4 6 Amps
Voltage 204 204 204 204 Volts

E Room: (103 + 23) * 198 * 1.732 = 43210 VA = 43.2 kW = 13 tons
W Room: (76 + 6) * 204 * 1.732 = 28973 VA = 29.0 kW = 9 tons

The East room obviously has cooling capacity to spare – as does the West room where there are six (6) 7.5-ton units installed and especially considering the more modern servers recently installed – but both lack the controls hardware and software capability required to provide reliable and energy efficient operations of the CRAC units. The lack of controls capability for the CRAC units has been and is a root cause of poor performance and high-energy consumption.

Previously, all units operated 24-hours/day, even though two (at least) in each Data Room are redundant. Unnecessarily operating all units was obviously wasting energy, but was also generating a fan energy cooling load of approximately 7,000 - 20,000 Btu/H per operating unit depending upon the fan motor brake horsepower. In other words, running all of the CRAC units was actually adding load to the Data Rooms due to the small actual cooling requirement and constant volume, in air stream fan/motor combinations.

The CRAC units were originally purchased with non-spring return chilled water valve actuators, and the on-board Liebert controls are programmed such that the chilled water valve actuators stay in their respective last position when the unit is de-energized. Because the chilled water system is primary-secondary pumped, and is variable flow secondary (using variable frequency drives), it is highly probable that the chilled water system was flowing much more chilled water than required – and wasting energy - when any of the units were staged off.

It is also worth noting that the condition of many of the CRAC units is poor. None of the integral infrared humidifiers were operable, many units have moderate to severe corrosion (rust), two of the units had inoperable chilled water valve actuators, and one had a faulty control board. The plenum space under floor is also quite dirty, and could possibly lead to premature failure of the computer hardware as the server racks are supplied with conditioned air from below the raised floor. Quotations for repairing the units, cleaning the units and the under-floor plenum, and establishing a preventative maintenance contract with a qualified service organization were provided.

Computer Room System Solutions:
Many iterations of the chilled water coils (Liebert CRAC units) provided sufficient information to formulate and recommend an integral central plant operating strategy. Further analysis of the CRAC’s indicates satisfactory operation of the server racks with approximately 63-degree supply air, which is achievable with approximately 60-degree entering water temperature (EWT) and design (52 GPM) flow rates As the supply air temperature from the CRAC units is paramount to reliable server and resultant simulator operations, the performance of these chilled water coils played a vital role in overall system operations, performance and energy efficiency.

To establish control of the CRAC units, a Liebert RAC-8 controller and temperature sensor were installed in each Data Room. The RCA-8’s enable the units to be duty cycled, manually over-ridden and started/stopped depending on space temperature and user requirements. Energy-Environment-Economics was also able to develop and implement a method for driving the chilled water valve actuators closed when the unit is off, enabling a lower DP setpoint for the secondary pump variable frequency drives.

The CRAC units’ modeling during the optimissioningSM phase established setpoints for staging the CRAC units’ on/off, based upon the room temperature, that will enable the CRAC the ability to maintain ample cooling capacity during both chiller and heat exchanger modes of operation. The duty cycle will rotate the lead CRAC unit on a 2-week basis. See Section for all Tables.

Air Handling System Deficiencies:
The air handling unit chilled water coil selections reflect a total chilled water flow requirement of approximately 209 GPM, indicating a design day load of approximately 87 tons was the engineered load point for the four (4) zones. Our load calculations indicate a required peak load of approximately 49 tons, which equates to approximately 118 GPM based upon a 10-degree water temperature rise and includes the packaged unit loads serving the inside of the simulators. Design flows to all Lobby fan coil units totaled approximately 150 GPM, or approximately 63 tons. Our load calculations show a design load of approximately 25 tons, or 60 GPM for a 10-degree rise coil.

Obviously there is a considerable amount of excess capacity, both water and airflow, available to serve the respective zone loads. Field measurements of the supply air temperatures were taken, with the checked air-handling units delivering approximately 62-degree supply air during typical “winter” conditions. This is important information for the development of new controls strategies to utilize the plate-frame heat exchanger and reduce overall energy consumption.

Our load profiles for the air handling unit zones indicates more than sufficient capacity, even considering that the simulators in the East and West bays were using nominal 4-ton packaged heat pumps for simulator (inside) cooling, and rejecting the heat directly into the bay. Considering the heat of compression, while operating these heat pumps are adding approximately 10 tons of load to the East bay, requiring the air handling unit(s) and ultimately the chilled water system to reject this heat load. The West bay, newly outfitted with simulators, also has two nominal four (4) ton packaged units rejecting loads directly to the control zone(s).

In general, the air handling unit’s original selections including airflow quantities, coil selections and control valve selections were less than ideal. These factors have contributed to the poor performance and high-energy costs of the complete HVAC system. For example, the air handling units’ inability to modulate the chilled water flow effectively has helped to create very low (approximately 4 degrees) system temperature differential (DT) and increased pump HP. Low system DT’s also has the ultimate effect of limiting the loading of the affected chiller(s), necessitating that additional chiller systems to be started to compensate.

Specifically, air handling unit #3, equipped with a variable frequency drive, was discovered to have an unnecessarily low supply air temperature reset setpoint (50-degrees). Coupled with it’s associated VAV boxes (apparently) never having been commissioned, the reset schedule – established to set the supply air temperature required to satisfy the greatest zone demand - was continuously calling for 50-degree air even during winter months, requiring that much more unavailable reheat. Nearly all of the VAV boxes were severely out of calibration, further exacerbating space temperature deviations from setpoint and subsequent tenant complaints.

Air Handling System Solutions:
The chilled water valves were replaced with pressure independent 2-way control valves to reduce the over-pumping problem. Experience with these valves has been excellent, and the infinitely variable Cv automatically compensates for fluctuations inherent in DP controlled variable secondary pumping systems. These valves, coupled with the central plant revisions and the new sequence of controls, have more than tripled the system temperature differential (DT). As evidenced by the DP setpoint (previously 12.5 psig, now at 3.75 psig) and one (two previously) secondary pump operating at approximately 40Hz (two at approximately 52 Hz previously), the chilled water system

Direct Digital Control (DDC) System Deficiencies:
The DDC system was a complete disaster. All of the necessary capabilities were resident, they simply had not been implemented into a comprehensive energy management strategy. Setpoints were unrealistic or too high, PID loops were not tuned, sensors were out of calibration, reset schedules were too aggressive, not aggressive enough or non existent.

Direct Digital Control (DDC) System Solutions
All existing sequences were reviewed and either rewritten entirely or modified. New programming was added, all setpoints were optimized, existing sensors were calibrated or replaced. Essentially, the entire DDC system was revamped.

Final Results
After commencing in November of 2004 and being released to full control March of 2005, the project has had an immediate and staggering impact on energy consumption and utility costs. Space comfort has improved, including the Data Rooms’ ability to maintain temperature and humidity with in specified limits. The project will conserve approximately 1 million KwH during the first (partial) year in fully automatic operation. The Kw demand has almost been cut in ½.

Considering that only the HVAC system was optimissionedSM, and that lighting and approximately 400 HP of hydraulic pumping (serving the simulators) capability was left untouched, these results are nothing short of spectacular. Due to the concerted efforts of the optimissioningSM team, and especially Mr. Jeff Seaton, approximately 2/3 of all HVAC equipment was turned off, thereby generating these savings, extending the life expectancy of the existing equipment, and reducing their associated maintenance requirements and costs.

The following charts were created from utility data provided by Mr. Seaton, and generated by his Utility Manager software:





      Copyright Energy Environment Economics - 2003