<HTML>
I agree with both Dan Nall's and Bruce Spurlock's comments regarding detailed
thermal balance loads calculations vs. the real needs of the energy simulation
community, and would like to comment further as to how an HVAC system actually
responds to building loads.
<P>First, let me comment on my qualifications and biases. In the
late '70s, I was one of the original DOE-2 authors, responsible for many
of the PLANT algorithms. After leaving LBNL in the early '80s, I
designed HVAC systems and controls for over a decade. Since the early
'90s I have worked closely with JJH&Associates (Jeff Hirsch), and wrote
or co-wrote much of the new code in the air-side and water-side portions
of DOE-2.2. With JJH, I am currently the lead author of a component-based
refrigeration/mechanical system module slated for the next major DOE-2
release.
<P>In a nutshell, an HVAC system responds to a building's thermal loads
via a thermostat that senses the deviation of zone temperature from setpoint.
Given this, a highly relevant question is, "How accurately does a thermostat
measure space temperature, and how seriously can the error in measurement
affect the heat extraction rate of the HVAC system?"
<P>In terms of accuracy, there are a number of factors that preclude temperature
measurements more accurate than 2-4°F, including:
<OL>
<LI>
Thermostat location - Energy simulation programs assume the thermostat
measures the air temperature, with a possible modification for radiant
flux. In reality, a thermostat measures a combination of air temperature,
radiant temperature, and wall temperature. The interplay between
these factors can be significant. For example, most residential thermostats
incorporate a small internal resistance heater that energizes whenever
the space heater is operating. Otherwise, the lag in sensed vs. actual
space temperature will result in excessive temperature overshoot and discomfort.
(Many thermostats allow you to adjust this "anticipator". If your
heater runs for long periods of time and overshoots the setpoint, you need
to reduce the anticipator's resistance, thereby increasing the electric
current and internal heating rate. If your heater short cycles, you
need to increase the resistance, thereby reducing the current and heating
rate.)</LI>
<LI>
Temperature stratification/gradients - Temperature gradients typically
exist between outer surfaces and inner surfaces, and between the floor
and ceiling. A VAV system in the heating mode can easily have a vertical
gradient of 10-20°F due to air stratification. (How many thermal
balance calculations recognize that VAV systems produce a hot ceiling surface
during heating, and a cold surface during cooling??) In the cooling
mode, an HVAC system typically has a gradient of 20-30°F between supply
and return air. The supply air does not instantaneously mix with
room air; instead it flows along the ceiling and down the walls as it gradually
mixes. A thermostat located in this partially-mixed regime may read
a temperature significantly different than the average room temperature.
Worse, the error is not constant, it varies as the supply air volume and
temperature varies. A CFD program could take these effects into account,
but would require the location of the thermostat to be explicitly identified,
as well as the shape and location of all furniture, as well as the location
and performance characteristics of all supply diffusers and return registers.
And of course, you would have to specify which interior doors are open
and which are closed! This level of detail is hardly practical in
most cases.</LI>
<LI>
Calibration - Electronic thermostats are significantly more accurate than
the old pneumatic ones, but may still be in error by 0.5-1.5°F.</LI>
<LI>
Thermostat wars - He's hot, she's cold. Ask any auditor how many
thermostats they find with broken "tamper proof" covers. Since the
goal of any HVAC system design is comfortable, productive people, it makes
sense that people be allowed to adjust their space temperature to match
their mood or their physiological needs. But that creates one more
unknown in the simulation model.</LI>
</OL>
So, if a thermostat may be off by 2-4°F from what you, the building
simulator, thinks is the setpoint, what error does this cause in the simulation?
The intent of a thermal balance calculation is to be able to calculate
the <I>instantaneous</I> temperature at any given time so that the <I>instantaneous</I>
HVAC load can be calculated. But, if a thermostat is off by 2°F
and has a throttling range of 4°F, then the error in the HVAC system's
<I>instantaneous</I> extraction rate may be 50%!!
<P>What this means is that it is impossible to determine the <I>instantaneous</I>
load at any given point in time. A simulation program really cannot
determine whether a building peaks at 3:30 p.m. or 4:00. The most
you can expect is that a program should be able to place the loads on the
HVAC system within an hour or so of when they actually occur. Numerous
studies have demonstrated that both thermal balance and weighting factor
solutions have this resolution. I am not aware of any study that
has demonstrated any real-world difference in accuracy between the two
methods. Even if the physical parameters could be exactly specified
and calculated, the error in the human factors could not.
<P>For these reasons, as well as the concerns enumerated by Dan and Bruce,
JJH and consultants have chosen to focus primarily on improved mechanical
system algorithms and user interfaces. We believe that improvement in these
areas are of the greatest value to the user community.
<BR>
<BR>
<BR>
<BR> </HTML>