Research Papers

Optimal Setpoints for HVAC Systems via Iterative Cooperative Neighbor Communication

[+] Author and Article Information
Matthew Elliott

Stingray Offshore Solutions,
8901 Jameel Rd., Ste 100,
Houston, TX 77040
e-mail: matt.elliott@gmail.com

Bryan P. Rasmussen

Associate Professor
Mechanical Engineering,
Texas A&M University,
3123 TAMU,
College Station, TX 77843-3123
e-mail: brasmussen@tamu.edu

1Corresponding author.

Contributed by the Dynamic Systems Division of ASME for publication in the JOURNAL OF DYNAMIC SYSTEMS, MEASUREMENT, AND CONTROL. Manuscript received July 31, 2013; final manuscript received June 7, 2014; published online August 28, 2014. Assoc. Editor: Prashant Mehta.

J. Dyn. Sys., Meas., Control 137(1), 011006 (Aug 28, 2014) (15 pages) Paper No: DS-13-1295; doi: 10.1115/1.4027887 History: Received July 31, 2013; Revised June 07, 2014

Heating, ventilation, and air conditioning systems in large buildings frequently feature a network topology wherein the outputs of each dynamic subsystem act as disturbances to other subsystems. The distributed optimization technique presented in this paper leverages this topology without requiring a centralized controller or widespread knowledge of the interaction dynamics between subsystems. Each subsystem's controller calculates an optimal steady state condition. The output corresponding to this condition is then communicated to downstream neighbors only. Similarly, each subsystem communicates to its upstream neighbors the predicted costs imposed by the neighbors' own calculated outputs. By judicious construction of the cost functions, all of the cost information is propagated through the network, allowing a Pareto optimal solution to be reached. The novelty of this approach is that communication between all plants is not necessary to achieve a global optimum. Since each optimizer does not require knowledge of its neighbors' dynamics, changes in one controller do not require changes to all controllers in the network. Proofs of convergence to Pareto optimality under certain conditions are presented, and convergence under the approach is demonstrated with a simulation example. The approach is also applied to a laboratory-based water chiller system; several experiments demonstrate the features of the approach and potential for energy savings.

Copyright © 2015 by ASME
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Fig. 1

Simplified schematic of a typical building HVAC system

Grahic Jump Location
Fig. 2

Schematic of a NC-OPT subsystem. The controller only communicates with subsystems that affect or are affected by the controlled plant.

Grahic Jump Location
Fig. 3

Example of a system that can be setup as an NC-OPT network. (a) Physical collection of neighboring rooms with controllers, and uncontrolled exogenous disturbance coming from outside temperature. (b) Subsystems displayed as NC-OPT network.

Grahic Jump Location
Fig. 4

Experimental system schematic

Grahic Jump Location
Fig. 5

Network topology schematic, showing the physical signals passing between subsystems. Signal names are the same as in Fig. 4 and Table 1.

Grahic Jump Location
Fig. 6

Subsystem 1 block diagram. Compressor and condenser use compressor rpm to control suction pressure and power consumption.

Grahic Jump Location
Fig. 7

Typical block diagram, evaporators (subsystems 2–4). The EEV is used to track a superheat setpoint.

Grahic Jump Location
Fig. 8

Typical block diagram, zones (subsystems 5–7). The water pump is used to control temperature drop and zone temperature.

Grahic Jump Location
Fig. 9

Simulation convergence at t = 0: Inputs converging to centralized optimal values over cmax iterations

Grahic Jump Location
Fig. 10

Centralized cost t = 0 converging to minimum over cmax iterations

Grahic Jump Location
Fig. 11

Zone 1 temperature change and zone temperature inputs for two tests with the same setpoint step change but with two different numbers of iterations per sample, compared to centralized optimal values compared

Grahic Jump Location
Fig. 12

Zone 1 actual temperatures for two tests with the same setpoint step change but with two different numbers of iterations per sampling period

Grahic Jump Location
Fig. 13

Estimated disturbance

Grahic Jump Location
Fig. 14

Changing input setpoints for (a) superheat and (b) temperature drop, in the presence of an increase in heat load disturbance. User-desired setpoints are also shown.

Grahic Jump Location
Fig. 15

(a) Compressor speed and (b) power consumption in the presence of an increase in heat load disturbance

Grahic Jump Location
Fig. 16

Tradeoff between setpoint tracking and power consumption during an increase in heat load disturbance, The actual temperature and the setpoint are shown. Additionally, the actual NC-OPT prediction and the predicted results if the control inputs remained unchanged are shown.

Grahic Jump Location
Fig. 17

(a) Pressure, (b) compressor speed, and (c) power comparisons for two different power setpoints

Grahic Jump Location
Fig. 18

Zone temperature tracking comparisons




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In