Appendix B. Diagnostic Plots
We have identified ten diagnostic plots that can be used to
analyze building and cooling system performance.
Whole Building Characteristics
building electrical power over time
Plot 1a. 3D carpet plot (power vs day of year vs time of day)
Plot 1b. 2D plot (power vs time of day) for all days of year
Plot 1c. 2D plot (power vs time of day) for weekdays only
whole building electrical power vs outdoor temperature
Plot 2a. All days of year
Plot 2b. Weekdays only
whole building electrical power vs outdoor temperature, for each 24 hour
Cooling System Characteristics
cooling system efficiency vs cooling load (tons)
system electrical power vs outdoor temperature
efficiency vs cooling load (tons)
efficiency over time
Cooling Tower Characteristics
Plot 7a. 3D carpet plot (efficiency vs day of year vs time of day)
Plot 7b. 2D plot (efficiency vs time of day) for all days of year
Plot 7c. 2D plot (efficiency vs time of day) for weekdays only
tower electrical power over time (This plot was not included in the
original list of nine plots.)
tower approach temperature vs cooling produced (tons)
tower cooling produced (tons) vs condenser flow
1. Whole Building Electrical
Power over Time
The plots above have been developed for a prototypical Northern California
office building. This building is based on 74 buildings and their average
energy use. The prototype is assumed to be located in the Alameda area
and has 100,000 square feet of floor space in five stories. The exterior
walls have R-22 insulation, and the roof has R-25 insulation. Windows are
single-pane, with an average shading coefficient of 0.43 and conductance
of about 1.0 Btu/hr/sqft/deg F. About 43% of the exterior wall area and
29% of the building wall and roof envelope is glass. The building is occupied
by 365 people (about 274 sqft/person). The maximum lighting demand is assumed
to be 1.455 W/sqft, and maximum electrical equipment demand is 0.705 W/sqft.
The building is assumed to be almost fully occupied, lighted and equipped
between 9am and 5pm weekdays, with tapered loads during night and weekend
hours. The building is assumed to generate positive pressure whenever the
perimeter fans are on, but allows infiltration of 0.07 cfm/sqft at all
The building is cooled by two electric-powered centrifugal chillers, using
two cooling towers with variable-speed fans as heat sinks. The full-load
efficiency of the chiller is 0.71 kW/ton. The building is heated by 2 natural
gas-fired hot water boilers with a heat input ratio of 1.33, hooked up
to a single-zone reheat system.
The building is heated to 69.6° F from 6am to 8pm on weekdays and from
9am to 7pm on weekends from November through February. The building perimeter
is cooled to 72.4°F for the rest of the year during the same hourly
periods. The building central core (about 57% of the total building floor
area) is cooled as needed 24 hours a day, every day. Likewise, the fans
run in perimeter areas during the weekday and weekend periods specified
above for heating, but the core area fans run all the time.
Whole building power is first normalized by dividing by the total floor
area of your building to get a Watts per square foot (W/sqft) value. You
can then examine this value over time in different ways to see how it varies.
Figure B–1a looks at an entire year's
worth of whole building power on a three-dimensional "carpet" plot. The
turquoise x-axis shows the day of the year, and the yellow z-axis shows
the hour of the day, while the vertical y-axis plots the value of whole
building power. This plot is able to show the variations in electricity
use during the year. From this plot you can make several checks. Are your
peak usage values at reasonable levels, and do they show up when you expect
them? Are there any unexpected peaks? How about your minimum electricity
usage? Does it correspond well with unoccupied hours? Can it be reduced
further, or kept lower for longer hours?
The prototypical Northern California Office Building data shown below is
an idealized version of a well-controlled building. Power use peaks slightly
during the summer (days 180 to 300), when electricity is used to run the
chillers, and tapers off in the winter. This annual profile would show
more variation in a harsher climate than the Northern California Bay Area's.
This building's peak usage of about 4.5 W/sqft occurs during summer afternoons.
The power use drops fairly cleanly to a minimum of 1.25 W/sqft at night.
There are no unexpected usage patterns in this prototypical 3-D carpet
Figure B-1a. Whole Building Power
Another way to look at the whole building electrical power use is to
use a simpler two-dimensional plot of power (y-axis) versus time of day
(x-axis). This plot can be viewed in conjunction with, or instead of, the
3-D carpet plot. In Figure B-1b below
you can see peak usage, baseline usage and the daily load shape more clearly.
You can also see clearly a number of low usage days along the bottom of
Figure B–1b. Whole Building Power
vs. Time of Day – All Days
It's useful to make 2-D plots of only weekdays, weekends, or a specific
day of the week to evaluate the its power profile. Figure
B–1c shows whole building power on weekdays only, but including
holidays. Notice that most of the low power days have dropped off the plot,
signifying that almost all occupied days use significant power. A plot
of weekends & holidays only would also be useful, in order to ensure
that power is kept to a minimum when the building is unoccupied.
Figure B–1c. Whole Building Power
vs. Time of Day – Weekdays
Whole Building Electric Power vs Drybulb Temperature
Looking at whole building power
versus the outdoor drybulb temperature is a good way to determine your
building's dependence on weather conditions. (Remember to normalize your
power use by floor area, to plot W/sqft). If your building is electrically
cooled but not electrically heated, as is the case with our prototypical
building, you should see your maximum power correspond to the hottest drybulb
temperatures. If you use electric heaters, you would expect a more U-shaped
curve, with high power use at both low temperatures and high temperatures.
In Figure B–2a, there are three data
"clumps". The lowest clump occurs when no chillers are running to cool
the building. The middle clump shows power use when one chiller is on,
and the highest clump shows power use when both chillers are in use. Note
the increasing slope of each clump, defining the increasing dependency
of power use on temperature as more cooling equipment is used. Watch out
for stray data points, or small clumps points in unexpected locations.
Figure B-2a. Whole Building Power
vs. Drybulb Temperature – All Days
One can also look at specific days, weekdays only, or weekends only.
Figure B–2b below of the Northern California
prototype looks at only weekdays. The most noticeable difference between
the weekday plot and the all day plot is there are fewer low-power use
Figure B-2b. Whole Building Power
vs. Drybulb Temperature – Weekdays
Whole Building Power vs Outdoor Temperature, for each 24 hour period
Another meaningful way to examine whole building power is to look at
each hour of the day separately. This way you can clearly see hourly trends
in power use, and quickly determine where your energy use might be lowered.
Figure B–3 below is a sample plot from
the prototypical Northern California building of power versus outdoor drybulb
temperature. There are a number of things to be observed here.
Figure B–3. Whole Building Power
vs. Drybulb Temperature – All Days
4. Cooling System Efficiency
versus Cooling Load
First, you can see that the minimum energy use occurs during the nighttime
and early morning hours, when the building is unoccupied. The flat nighttime
load shape shows that little if any cooling is used at night. During hours
5 and 6 you can see power use separate into two lines, the lower line for
unoccupied days and the higher for occupied days. The power increases during
hours 5 and 6 occur as lights and appliances begin to come on in the building.
Note that the power use of these non-cooling related appliances is generally
not sensitive to outdoor temperature, as shown by their non-sloping appearance.
At hour 7, power use not only rises further, but starts to become temperature
dependent. This signals that fans and/or chillers are now used to cool
the building. Also at hour 7, the lower unoccupied line begins to split
into 2 separate lines as the building core area may need some cooling on
some of the minimally occupied days.
Through the morning and into the afternoon, the power versus temperature
curves spread further apart as the power used for lights, appliances, cooling
and ventilation peaks. At hour 15 the power use begins to taper off, and
the curves slowly coalesce as the end of the day is reached.
A plot of your cooling system efficiency versus the delivered load
is a useful tool for system analysis. This plot can be compared to theoretical
efficiency curves for your system. It can also be monitored over time to
check for degradation of your cooling system performance. Cooling system
performance is dominated by the chiller, so the shape of this curve should
look a lot like your chiller performance curve. Overall system efficiency
should lie somewhere between a low 2.0 kW/ton and a high 0.7 kW/ton, with
newer systems having higher efficiencies. Figure
B–4 shows the efficiency curve for our prototypical building. This
idealized curve is very well defined. An actual curve will have more scattering
of points. But every curve will have an area where the curve generally
flattens out and the optimal efficiencies are reached. The curve below
flattens out above 30% of the maximum load. To maximize system efficiency
you want to run your system at these loads as much as possible, and avoid
the lower-load, steeper part of the curve where the system is less efficient.
Figure B–4. Cooling System
Efficiency vs. Cooling Delivered
Comparison of your efficiency curve to system specifications
or to historical performance of your system can be used to find many problems.
Your efficiency curve may not be as high as the specs, or as high as it
was a year ago. This may be caused by a number of things, including:
Cooling System Electrical Power versus Outdoor Drybulb
System changes (such as cooling tower
Poor water flow characteristics (perhaps you need multiple chillers)
Component malfunctions (like condenser fan cycling)
Fouling of chiller tubes
Loss of refrigerant charge
Poor full-load or part-load performance (may be related to weather conditions)
Over- or under-sizing of components
Loss of efficiency is not the only thing to look for on your cooling
system efficiency curve. You also want to look at where you're operating
on this curve, and for how long. Watch out for:
Excessive on time
Heavy system use at low efficiencies.
Another interesting plot is cooling system power versus the hourly
outdoor drybulb temperature. This curve shows the temperature dependence,
with power use increasing with outdoor temperature. Figure
B–5 below, for our prototypical Northern California office building,
also clearly shows two lines of system operation. The lower line shows
operation with only one chiller in operation, and the upper line is when
both chillers are running.
Cooling system power is highly dependent on outdoor temperature. But
it also depends on building loads, such as solar loads, lighting and appliance
use, and people. In the prototypical plot below there is quite a bit of
load scatter at any given outdoor temperature. In general, the load is
met by a single chiller at low outdoor temperatures, and by both chillers
at high outdoor temperatures. But at non-extreme temperatures one or both
chillers may be needed to meet the building loads.
Figure B–5. Cooling System
Power vs. Drybulb Temperature – All Days
Chiller Efficiency versus Chiller Cooling Load
It is valuable to track of the efficiency of your chillers, either
through continuous monitoring or by making periodic checks of performance
for various loads. It is useful to compare their performance to the manufacturer’s
specifications, as well as to track performance over the chiller's lifetime.
Figure B–6 shows chiller efficiency in
kW/ton versus the cooling load in tons for our prototypical Northern California
office building. Note that this curve reaches an optimum efficiency plateau
at a load of about 100 tons. For best overall performance, you want to
operate your chiller at or above this level. The newest, most efficient
chillers plateau at full-load efficiencies of 0.6 kW/ton or better. Older,
less efficient chillers may have full-load ratings of 0.9 to 1.0 kW/ton
Figure B–6. Chiller Efficiency
vs. Chiller Cooling Produced
Loss of efficiency over time may be caused by:
7. Chiller Efficiency
Poor water flow through the chiller
Fouling of the chiller
Lack of refrigerant charge
Poor efficiency on part of the curve may be due to low loading conditions
during mild weather
In addition, watch out for how often your chiller is being used, and
where that usage is on the efficiency curve:
Excessive on time at full-load
Chiller may be undersized
Delivery system may not be working properly
Setpoints or control schedules
may not be optimal
Frequent cycling behavior and/or heavy part-load use
Chiller may be oversized
Temperature gauges or thermostats may
Control schedules may not be
Looking at chiller efficiency over time will help you to see if you're
using your chiller effectively. Efficiency can be plotted on a three-dimensional
carpet plot (efficiency vs day of year vs time of day) or on a simpler
two-dimensional plot of efficiency vs time of day. Either plot can be used
to pinpoint where chiller efficiencies are lower than expected. These plots
can be used to fine tune your operating schedules, or to trouble shoot
for chiller problems.
Figure B–7a shows the 3-D carpet plot
of efficiency for our prototypical Northern California office building.
You can see the best efficiency is achieved in the middle of the day when
the chiller is running at full load. Seasonal variations can also be observed.
The full load "channel" is wider during the summer months, when chiller
loads are higher for a longer part of the day. Part-load efficiencies are
also better on the "shoulders" during the summer months. During the winter,
shorter periods of full-load operation and longer periods of less efficient,
part-load operation are observed.
Figure B–7a. Chiller Efficiency
The 2-D plot of the prototypical Northern California office building is
shown in Figure B–7b for all days of operation
for a year, then in Figure B–7c for weekdays
only. In these two plots you can see that weekend/weekday operation is
slightly different. On weekdays chiller operation begins at 5 or 6 am,
while on weekends operation doesn't begin until 8 am. Weekends are less
likely to reach the most efficient full-load conditions, since only the
core of the building is being cooled and there are lower building loads.
Figure B–7b. Chiller Efficiency
vs. Time of Day –All Days
Figure B–7c. Chiller Efficiency
vs. Time of Day – Weekdays
Cooling Tower Power over Time
Looking at cooling tower power over time will help you to see when
you're using your cooling tower. This plot can be a three-dimensional carpet
plot (power vs day of year vs time of day) or a simpler two-dimensional
plot of power vs time of day. These plots can be used to fine tune your
operating schedules, or to trouble shoot for cooling tower problems.
Figure B–8 shows the 2-D plot of cooling
tower power for all days of the year in our prototypical Northern California
office building. Looking at this plot together with plots of chiller power
use can ensure that the cooling tower is always working when the chiller
is running. The prototypical cooling tower for our building runs at two
flow levels, one using about 0.8 kilowatts over a full hour of run time
and the other using about 1.6 kilowatts to pump double the flow level.
If the cooling tower was undersized, as is common, this plot would show
significantly more operation time at the 1.8 kilowatt level.
An undersized cooling tower can seriously impact the chiller efficiencies!
A too-small cooling tower will deliver warmer than necessary water to the
condenser, not allowing the chiller to release as much heat and worsening
the chiller's efficiency. This problem will degrade chiller operations
when you most need optimal performance - on hot days and during afternoon
load and energy-rate peaks.
Figure B–8. Cooling Tower Power
vs. Time of Day – All Days
Cooling Tower Approach Temperature vs Cooling Produced
Another useful plot for evaluating cooling tower performance is the
cooling tower approach temperature versus the tons of cooling produced.
The approach temperature is defined as the condenser water supply (or cooling
tower exit) temperature minus the wet bulb temperature. As more energy
is extracted from the cooling tower flows, the condenser water supply temperature
"approaches" its limit of the wet bulb temperature, and the approach temperature
Figure B–9 is a 2-D plot of cooling tower
approach versus the cooling tower cooling produced for all days of the
year in our prototypical Northern California office building. The approach
temperature is reduced as more cooling is produced. This plot can be watched
to see if the approach temperature is increasing over time at different
tonnage levels - an indication of fouling of the cooling tower and a signal
to clean it. More efficient cooling tower operation can deliver cooler
water to the condenser of the chiller, for better chiller efficiencies
and lower energy use.
Figure B–9. Cooling Tower Approach
vs. Cooling Produces – All Days
Tower Cooling Produced vs Condenser Flow
A plot of cooling produced versus condenser flow allows you to keep
track of the amount of flow through the cooling tower and how much cooling
is produced from the cooling tower alone.
Figure B–10 is a 2-D plot of cooling
tower cooling produced versus the condenser flow for all days of the year
in our prototypical Northern California office building. This theoretical
cooling tower runs at two flow levels, 190 or 380 gallons per minute. Since
this is a theoretical building, the condenser flows remain exactly the
same all year, with no flow reduction due to fouling of the piping or the
cooling tower, and no piping leaks.
In an actual building, you must watch for flow reductions over time
as the cooling tower becomes clogged and dirty. You must also watch for
higher than normal flows through the cooling tower, which could indicate
problems with pump operations.
Figure B–10. Cooling
Tower Cooling Produced vs. Condenser Flow – All Days
Start of Report
Section 1. Project Overview
Section 2. Pilot Site Selection and Technology
Section 3. IMDS Description and Accuracy
Section 4. Building Performance and Findings
from the IMDS
Section 5. Automation of Diagnostics
Section 6. Economic Issues and Related Technology
Section 7. Conclusions and Future Plans
Section 8. References
Appendix A. Web-Based Performance Analysis Tools
Appendix B. Diagnostic Plots
Appendix C. IMDS Points, Sensors, and Data
Appendix D. IMDS Findings Report Log
| Commercial Building Systems
| Building Technologies
This page last updated 7/10/99 by SKhalsa