Vous êtes sur la page 1sur 5

Short Report

Initial experimental evaluation of alternative control valves for VAV


reheat

Submitted to

Brian Phillips

Belimo Americas

33 Turner Road

Danbury CT 06810

Email: Brian.Phillips@us.belimo.com

Submitted By

Jie Cai, Graduate Student

James E. Braun, Primary Investigator

Orkan Kurtulus, Post-Doc

Purdue University

School of Mechanical Engineering

Ray W. Herrick Laboratories

177 S. Russell St.

West Lafayette, IN 47907

August 18, 2015


I. INTRODUCTION
This project aims to evaluate and study the performance of two different types of control valves having
different characteristics for a case study system. Comparative testing of the valves is accomplished for
reheat coils that serve graduate student offices that are living laboratories within the Center for High
Performance Buildings (CHPB) at Purdue. The Living Labs (LL) are four nearly identical office spaces.
Each office is reconfigurable in different ways that can enable direct comparisons of alternative
technologies for windows, lighting comfort delivery, control and acoustic treatments. Comfort delivery
options include air supply from the ceiling, floor or side wall along the radiant floor heating and radiant
chilled beam cooling.

The project was motivated by poor comfort control performance observed in the HVAC systems serving
the Living Labs, which was believed to be caused by significant valve hysteresis for both the chilled- and
hot-water valves. Conventional PID control was not able to provide stable control due to valve hysteresis
and highly non-linear behavior. The control variables were observed to oscillate significantly.

This report presents some initial experimental test results on the performance of valves from two different
manufacturers. The original valve, termed "old valve" in this report, was specified in the building design
and installed during the construction. The second valve type from Belimo was donated for this project
(termed "new valve" in this report) and was installed as a retrofit for reheat valves in LL#3.

Figure 1 shows the air distribution system layout of LL#3 at CHPB. The system has two air distribution
modes: overhead air distribution (OAD) mode and underfloor air distribution mode (UAD). The OAD
system has three VAV boxes and the UAD system has two VAV boxes. Each VAV box has an air
damper that modulates airflow rate. During the testing period the airflow was maintained at the minimum
level in order to enable the reheat. The old valves were installed in the OAD VAV reheat control during
the building construction phase, while the new valves were installed as a retrofit for the UAD VAV
boxes. In the preliminary tests, only LL#3 was considered and only one air distribution mode could be
enabled at a time.

OAD VAV
w/ old valve

Office Space

UAD VAV
w/ new valve

Figure 1. Air distribution system layout of the test room. LL#3 (#3057)

Initial tests were designed to evaluate the linearity and hysteresis effects associated with the valves by
sweeping the valves through their operation range and evaluating the heating capacity response of the
reheat coils. Then closed-loop control tests for maintaining occupied space temperature were performed
in LL#3 with the OAD and then the UAD to evaluate the control behavior of the valves.
II. LINEARITY AND HYSTERESIS EFFECT TESTS
In these linearity and hysteresis effect tests, the valve control signal was stepped from 0% to 100% and
then back to 0% with a 5% increment and 8 minutes hold time for each step. The overall tests spanned 5.5
hours each and were carried out on summer days in LL#3 from 1:00am to 6:30am when the lab area was
unoccupied. Figure 2 shows the behavior of the old and new valves from the tests. The old valve data
was collected for the overhead VAV with new valve data associated with the underfloor VAV. The x-axis
is the control signal that was fed to the valves and the y-axis is the temperature increase across the reheat
coil in F. Since the airflow rate was maintained constant, the y-axis data provides a direct indication of
heating rate of the reheat coil.
Old Valve (B) New Valve (D)
60 40

35
50

30

40
25 Inc
Del T VAV (F)

Del T VAV (F)


Dec
Inc
30 20
Dec

15
20

10

10
5

0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Control signal (%) Control signal (%)

Figure 2. Open-loop test results for the old and new valves

A. Control continuity and linearity


It can be observed that when the valve was opening from fully closed, the old valve provided almost no
capacity when the control signal was below 20%. Once the control signal reached 20%, there was a
significant jump in the heating rate to almost 50% of the maximum heating capacity. This discontinuity in
the control effect causes severe problems in the control stability when the reheat demand is moderate
since effective control is lost in the capacity range of 0 to 50%. When the valve was closing from a fully
open position, a similar significant drop occurred when the control signal reached 10%. The new valve,
however, had a much smoother behavior over the whole testing range.

Note that both new and old valves have equal percentage control so the heating capacity should be
relatively linear with control command. However, the old valve exhibits significant nonlinearity while
good linearity is observed with the new valve.
B. Hysteresis
The old valve exhibits hysteresis in the low operating range, i.e., 5% - 40% with a very significant
hysteresis occuring at 15% control signal. Hysteresis can lead to poor control performance in these
aspects: (1) there is always a delay in the control action that slows down the control effect; (2) chattering
can occur when control is aggressive. The combination of hysteresis and non-smooth behavior could lead
to significant chattering. The new valve only shows less hysteresis within a smaller range, 5% - 25%,
which should provide better control performance compared to the old valve. Note that hysteresis is
inevitable as long as a gear train is used. But a better manufactured gear box can reduce hysteresis effects
to a tolerable level for a building application.
III. CLOSED-LOOP CONTROL TESTS
In the closed-loop tests, the system operated under its default control logic where the VAV box reheat
valve was modulated to maintain the space temperature at the setpoint with the minimum air flow setting.
Figure 3 shows results for the new and old valves obtained through sequential testing over a number of
days. The UAD mode was enabled for the first four days as a test of the new valves while the subsequent
4 days utilized the OAD mode for the old valve testing. The space temperature setpoint was set to a fixed
value of 72.1 F. It can be seen from the plot that the new valve was able to regulate the space temperature
within a tighter band around the setpoint than the old valve. The old valve led to significant fluctuations
in the space temperature due to its non-smooth behavior and significant hysteresis. In addition, due to the
nonlinearity in the old valve performance curve, the PI control output range of the old valve was much
smaller than that of the new valve. Figure 4 shows scatter plots of the temperature rise across the reheat
coils versus the valve control signal for both old and new valves. The reheat capacity with the new valve
has a clear correlation with control signal while the old valve data exhibits a nearly random and
uncorrelated behavior due to the chattering.

73
NewValve
72.5 OldValve
T zone (F)

72

71.5

71
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000

40 NewValve(D)
Control signal (%)

OldValve(B)
30

20

10

0
1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000
Time (min)

Figure 3. Closed-loop testing results with new and old valves tested in series.

New Valve (D) Old Valve (B)


30 30

25 25

20 20
Del T VAV (F)

Del T VAV (F)

15 15

10 10

5 5

0 0
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
Control signal (%) Control signal (%)

Figure 4. Valve characteristic comparison under closed-loop operation


IV. CONCLUSIONS
This short report presented some initial performance comparisons of VAV reheat valves from two
different manufacturers. From linearity and hysteresis test data, significant performance differences were
observed in the tested valves in regard to control smoothness, linearity and hysteresis. In addition, the
new valve performed much better than the old valve for closed-loop feedback control of LL#3.

V. NEXT STEPS
The comparisons of the two valves for feedback control of LL#3 are less than ideal because the dynamic
response of the UAD system could be slower than the OAD system. We would like to perform a more
direct comparison of the closed-loop performance for the same OAD system. Since we already have
characterized the performance of the OAD for LL#3 with the old valves, we would like to change those
valves to the new valves and report the tests. This will require an additional set of valves from Belimo.

Vous aimerez peut-être aussi