The Instability of VAV Systems

 Reprinted from Heating/Piping/Air Conditioning February 1992, Used with permission.

by Gil Avery

How control loop interaction can cause instability in variable air volume systems

An industrial instrumentation control engineer would shudder if asked to control the typical variable volume air handler, shown in Fig. 1, with the control hardware the HVAC industry is accustomed to using and with the budget he would have. And yet engineers continue to specify these systems even though there are few, if any, that operate as intended.
This paper only addresses the problems associated with large, typically field-assembled, VAV systems that incorporate at least the following components: an air handling unit with a supply fan that has a means of controlling the air volume, a return air fan that tracks the supply fan using air flow measuring stations, pressure independent VAV terminals, and an outside air economizer assemblage (Fig.1).

Control Loop
This system has four or more distinct control loops. Control loop SP-1 (shown in Fig. 2) consists of a duct static pressure sensor that controls the duct pressure through a variable frequency drive on the supply fan. 

The amount of air supplied to each zone is controlled by a room thermostat that resets a velocity sensing device in each VAV terminal. This device, in turn, operates the VAV terminal damper actuator to vary the volume of air supplied to the zone. Many VAV terminals may be supplied by one air handling unit and therefore may have many T-2 control loops (Fig. 3). 

An outside air economizer cycle (Fig. 4) is provided to allow the use of outside air for cooling when its heat content is low enough. Control loop T-1 with thermostat T-1 in the mixed air operates the return, relief and maximum outside air damper actuators.

Control loop FS-1 (Fig. 5) has a flow measuring station FM-1 in the supply air duct and a station FM-2 in the return air duct. Both are connected to a flow synchronizer, FS-1. The flow synchronizer varies the return fan speed so as to maintain the return air volume the same as the supply air volume less a fixed CFM. This CFM should handle the building exhaust, pressurization, and ventilation requirements. 

When the building HVAC system is commissioned, all four control loops can be shown to operate satisfactorily. To do this, one must disable the loops. For example, the return air flow can be demonstrated to correctly track the supply air flow using the sensors and transducers comprising control loop FS-1 (Fig. 5), if control loops SP-1 and T-1 are disabled. It would be difficult to demonstrate the performance of loop FS-1 if the static pressure sensor SP-1 was changing the speed of the supply fan and the mixed air temperature sensor T-1 was changing the position of the economizer dampers. These loops would both have to be disabled before the operation of control loop FS-1 could be demonstrated. The same procedure would have to be followed when demonstrating control loop SP-1, or mixed air control loop T-1.

It is only when all control loops are enabled that instability caused by control loop interaction becomes evident. For example, if the temperature in Zone 4 (Fig.1) drops, the damper on the terminal supplying this zone will start to close. When this occurs the supply duct pressure will increase, thereby increasing the flow through the other terminals. The pressure independent controls will react to close their dampers and the duct pressure will increase more. The static pressure sensor SP-1 will then lower the duct static pressure by reducing the speed of the supply fan. When this happens the fan tracking loop FS-1 will reduce the speed of the return fan so that the return air flow is reduced. When this reduction occurs, the mixed air temperature changes. Thermostat T-1 will then change the economizer damper positions. When the dampers change position, the pressure in the mixed air plenum changes, thus changing the pressure in the supply duct and the flow through all VAV terminals. This instability caused by loop interaction continues unabated. Very little can be done to eliminate the instability. PID algorithms do little to help because the interference caused by another loop is not a predictable upset. Loop feed forward software and hardware to correct this interaction is probably beyond the scope of our industry.

Gusty winds have a drastic effect on outside air intake and relief louvers. A forty-five mile an hour wind gust against a louver is equivalent to an air pressure change of 1" WG. This can wreak havoc with the mixed air thermostat and duct pressure controls. In many installations, the gusty wind pressure is great enough to blow into the relief louver and on into the mixed air plenum.

All four of the control loops incorporate either pressure or temperature sensors with very fast response times. The supply fan changes speed almost immediately with a change in duct static pressure. The return fan responds immediately to changes in duct air velocity sensed by the air flow measuring stations. The mixed air thermostat responds almost instantly to a change in temperature resulting from a change in flow of outside and return air. This changes the economizer damper position thereby instantly changing the system pressure.

Each pressure independent VAV terminal responds immediately to changes in terminal air velocity caused by duct pressure changes. Installations having a large terminal located close to a small terminal experience major changes in air flow in the small terminal caused by minor changes in air flow by the large terminal.

The Problems

Control loop interactions are difficult to pinpoint. Interaction between VAV terminals can occur spontaneously at any time or they may occur on a repetitive schedule every day. Frequently the interaction occurs after the air handlers have been in operation long enough for the terminals to all start throttling back. Interaction generally starts with the terminals near the end of the supply duct. Here the volume of air in the supply duct is the smallest and therefore, any change in VAV terminal volume will have the greatest effect on adjacent terminals, and on the supply duct static pressure sensor.

Identical systems may have different control loop interaction problems if the controls are not manufactured by the same company. This may occur because the valve and damper actuator stroke speeds may be different or the sensor response times may not be the same for different manufacturers. Damper and valve actuator speeds have a substantial effect on loop interaction. Pneumatic actuators usually operate faster than small bi-directional DDC electric actuators. Therefore loop interaction between pressure independent VAV terminals with pneumatic actuators may be greater than terminals with slower moving electric actuators.

Loop interaction between pressure independent pneumatic VAV terminals and supply fan electric inlet vane actuators can be severe, since the slow operating inlet vane actuator can't keep up with the changes in duct static pressure caused by the faster acting VAV actuators. This continuous hunting of the inlet vane controls will shorten the life of the vane actuator.

Most of the control loops were developed to solve a discreet problem. Supply-return fan tracking systems became popular because engineers thought it would solve the building ventilation and pressurization problem. Little thought was given to the impact this control loop would have on the other control loops in the system. Pressure independent terminals supposedly solved the perceived problem that changes in the duct static pressure would have on the zone air flow. Again, no consideration was given to the effect that this type of terminal would have on the other terminals or on the other control loops. The outside air economizer cycle has been used for years with good success on constant volume air handlers. But when used with the other control loops found in VAV air handlers, the performance has been disastrous.

The world's largest landlord, the U.S. Navy, has made an extensive study of this problem. As a result, the Navy is revising their VAV system specifications so as to eliminate much or all of the control loop interaction that was inherent with the old designs.

The indiscriminate use of pressure independent VAV terminals and the failure of control and design engineers to recognize the control loop interaction problems caused by the fan tracking systems, has devastated this segment of the HVAC industry. Many fan tracking control loops have been abandoned in favor of controlling the building pressure by varying the volume of air handled by the return air fan. This change does eliminate one of the interacting control loops, thereby improving system stability, but a constant amount of ventilation air is no longer supplied to the conditioned space. The amount of ventilation air will float with the supply fan CFM.

This change has surely contributed to some of the industry's woes with sick buildings that have poor indoor air quality. The publicity attributed to indoor air quality is growing every year. As a result, the engineer must be able to justify every phase of his or her design. There is no reliable technical information available to aid the design engineer in sizing economizer dampers for VAV systems that utilize return fans. Any engineer faced with a legal liability because of a sick building would find himself in an embarrassing position if he was questioned by the prosecuting attorney about sizing the ventilation dampers. An admission that there was no technology available for damper selection or that he left the selection up to the control contractor, certainly would not help his case.

Fan tracking VAV systems depend on the difference between the supply and return air flow sensor for the ventilation air. This difference is based on the accuracy of the flow sensing hardware. For example, if the ventilation requirements are 15% outside air and the accuracy of the supply and return flow sensors is ±5% of design, then under the worst case conditions, the ventilation air could vary ±60%. This is not acceptable.

ASHRAE Standards 62-1989 states that "when mechanical ventilation is used, provision for air flow measurement should be included." This provision is not normally part of the conventional fan tracking system. To add a sensor for reading ventilation air can be very expensive. The sensor must be capable of reading very low air velocities (50 FPM or less) and the sensors must be large enough to accurately sense the total flow in the 100% outside air intake duct when only 15% is flowing. 

The Solutions

Control loops for VAV air handlers can be configured to minimize interaction as shown in Fig. 6 and as described in the articles listed in the bibliography. The selection and sizing of dampers is also addressed.

Most commercial buildings do not require pressure independent VAV terminals nor can their additional cost and maintenance justify their use. Using correctly sized pressure dependent terminals will largely eliminate one interacting control loop.

The use of a relief fan rather than a flow tracking return fan will eliminate the most troublesome control loop and will also enable the engineer to properly size and characterize the outside and return dampers so that changes in damper position will not change the mixed air plenum pressure. This method eliminates the interaction between the economizer damper position and the supply duct static pressure sensor.

Properly sizing and characterizing the economizer dampers and using a relief fan will also minimize the gusty wind effect on the outside air intake and relief air openings.

A small outside air injection fan (sized to handle only the ventilation and pressurization requirements) will guarantee these requirements under all supply fan flow conditions.
 

Bibliography 

Alley, R.L., "Selecting and Sizing Outside and Return Air Dampers for VAV Economizer Systems," ASHRAE Symopsium Paper DA-88-18-1.

Avery,G. "The Myth of Pressure Independent VAV Systems," ASHRAE Journal, August 1989.

Avery,G., "VAV Economizer Cycle - Don't Use a Return Fan." Heating/Piping/Air Conditioning, August 1984.

Avery, G., "Updating the VAV Outside Air Economizer Controls," ASHRAE Journal, April 1989.

Engineering Investigation of Variable Air Volume HVAC Systems, U.S. Navy Contract N62467-87-C-0689.

Kettler,J., "Problems Associated With Return Fans on VAV Systems," ASHRAE Symposium Paper DA-88-18-3