Dear headsetsdotcom,
In our joint effort here, what we are designing is a system to
control heating .. (HVAC) in a residential home in California,
house with solar panels, pool and double-rate electricity.
We want to know the optimum time to
1) turn hot water heater on/off
2) turn the air cond/heating on /off
3) run solar to heat the pool
and to find devices that are intelligent enough to do that .
In other words, we want to design a smart HVAC control system.
First, we need to agree on the use of terms, so here are
some terms used in control engineering:
-----------------------------------------------------------------------------
Search Terms
HVAC smart control
HVAC smart control optimal
Closed vs Open loop control
The closed loop system is responding to the actual situation
as determined by sensors. Sometime the sky is cloudy and
high winds cool the pool. The average temperature
for that time of the day and day of the year , which an open
loop system would use would be a poor guidance in such a case.
Smart system
An intelligent system anticipates the response of the system as
well as the requirements. For example, If I usually come home
at 6pm and want temperature to be 70 degrees, the system needs
to know how long it takes to raise the temperature when heater
is turned on.
Optimal system
To optimise performance, the system has a concept of a goal.
Usually there are several goals: keep the living space and pool at
the selected
temperatures AND minimise the energy cost. The multiple goals
are combined into a Generalised Cost function C, using weights.
For example
C = Sum-over-time ( price * power + weight_1* (Ta1- Ts1) +
weight_2*(Ta2 -Ts2))
All quantities are a function of time
price(t) is given by the energy rates 8c/min per kw up to noon and
31c/min per kw from noon to 6pm
power(t) is the energy used by the system at that time, to achieve
the goal
weights w_n(t) are selected bu the user :
We can have several spaces, where we select what conditions we
want, when,
and how important that contion is:
For example:
We want temperature to be Ts (
T_selected ) at space 1,
the actual (measured) temperarure will be Ta
(T_actual)
By selecting weight_1 we are specifying what
we are willing to pay
to keep temperature near the selected value Ts
( e.g, w_1 = .01 $/ min * degree )
Weight w_n ai also function of time: If we are not home until 6pm,
the weight would be 0 until then.
The cost function can be more complex than in this example.
Weight can be zero for a deviation up to three degrees and rise
sharply after that...
The 'spaces and conditions may be' e.g temperature of the pool,
of the ID hot water, ..
Adaptive system
An adaptive system is able to learn from experience. To achieve the
optimum
(lowest generalised cost C) the system has to anticipate the
response of the 'plant' (see below).
It has to know e.g. the Thermal inertia of the pool, house, ..
and have statistics of
such random, unpredicable parameters as the weather.
Sometimes a fixed model of the controlled system (called the
"plant") is build into the control system,
but when the system changes from time to time, adaptive control
improves the model of the
'plant' as the system is learning.
These are the control system terms.
_________________________________________
Using these terms, our question can now be stated as follows:
_____________________________________________________
How can we design a smart adaptive closed control system
for a
residential object, with solar heating and pool, which will optimise
the General Cost function?
These problems, or ones like this, are being solved for the large
industrial plants,
http://www.bchydro.com/business/investigate/investigate854.html
The solution is provided by a computer, remote sensors, and a
computer program
written or adapted to a particular building complex by HVAC engineers.
The goal of our search is to find a solution which can be
implemented using a home computer
and that is smart enough to adapt to a specific home
automatically, possibly using AI algoritms.
(Thus eliminating labor-intensive engineering design specifc to a
particular plant.)
Search terms
://www.google.com/linux?hl=en&lr=lang_en&ie=ISO-8859-1&q=HVAC+control
1) The closest thing I have found to such a system is
NIc-aerc project (ID -- InDoors OD=OutDoors)
This project gets weather prediction data over the internet.
Predicted and measured OD temperatures are correlated to establish
a bias due to the micro-climate at the site,
Measured ID and OD temperatures are correlated to determine the
thermal lag time,
HVAC controls are correlated with measured ID temperature to monitor
the response rate of the house (i.e. based on HVAC capacity, on-site
weather, etc.)
Variable-rate energy pricing and the setpoint schedule are correlated
to develop
a cost-efficient HVAC schedule.
http://www.stratis-ltd.com/gpl/nic-aerc/index.html
It illustrates good design and provides interesting links to
resources
http://www.stratis-ltd.com/gpl/nic-aerc/refs_links.html
So, it can be done. The author here may be willing to share what he
has developed.
2) The idea of using the weather forcast was developed
by the Swedish branch of Honeywell into a product:
WeatherGain
When you use WeatherGain, the building's control equipment knows what
the
weather and the energy requirement will be, hour by hour, 3-5 days in
advance.
WeatherGain has been developed by Honeywell
INUcontrol and SMHI (the Swedish Meteorological and Hydrological
Institute),
where Professor Roger Taessler has calculated something called
" the equivalent temperature" (ET). The equivalent
temperature describes the effect of temperature, sun and wind,
in combination with data on the type of building, in terms of
location, orientation,
characteristics and function ...
http://www.honeywell.se/inu/
______________________________________________
Devices and protocols which communicate sensor data
_______________________________________________
To put a system together, we would start with some type of simple
home automation program
that we can modify and adapt. We are now searching for Open Source
resources.
Here are few components which could be used as seeds:
1) "MisterHouse is an open source home automation program.
It's fun, it's free, and it's entirely geeky. Written in Perl, it
fires events based on time,
web, socket, voice, and serial data. It currently runs on Windows
95/98/NT/2k/XP
and on most Unix based platforms, including Linux and Mac OSX.
It can talk, it can check your messages, control the lights, program
your VCR,
and what is best - it understands spoken commands."
http://www.misterhouse.net/
2) Here are some techniques for Automating Your Home
An enormously cool powerline protocol called X-10 enables compatible
devices
throughout the home to exchange information using the 110 volt wiring
that's already installed. Using X-10 products, you
can automate a heating/cooling system, create low-cost but advanced
security systems,
turn appliances on or off, control your home theatre, and much more.
(For a nifty directory of home automation Web
sites, see http://home-automation.org/.)
http://SoftwareForHomes.com/heating.htm
3) Here is a list of Linux-based home-control resources .
Some can be used on Windows systems as well, being based
on perls, tcl or other such OS independent language.
Also, the final control system could be run by an old computer
on which
linux with approprite software was installed.
http://www.geocities.com/linuxhomeautomation/
4) Here, one example from the above about swimming pool control
with Linux .
It deals with pool chemistry, not heating in this case, but
perhaps
it's is of interest
http://www.truetex.com/poolcontrol.htm
_____________________________________________
Your type of question is very current, globally speaking.
For background information there are many academic
research projects seeking solutions.
______________________________________________
Ongoing research, looking for such (AI) solutions
in Europe
......................
The EDIFICIO project, funded in part by
the EU
Commission in the frame of the JOULE III Programme, has the goal of
developing
innovative, adaptive, integrated control systems for the optimal
energy management
and indoor comfort in buildings. This is achieved by using
Soft Computing Techniques (SCT), specially Fuzzy Logic (FL),
Artificial Neural Networks (ANNs) and Genetic Algorithms (GAs).
http://www.csem.ch/detailed/pdf/e_Paper_Edificio.pdf.
... procedures and guidance on ventilation and source control. ...
materials, HVAC components,
complete HVAC systems and ... year-round performance of ?smart?
systems. ...
www.enerbuild.net/docs/PLEA_2000_Cambridge.doc
and in the US:
................................
.... We are fostering the development of more intelligent,
integrated, and optimized building
mechanical systems. We use a dynamic building heating, ventilating,
and
air-conditioning (HVAC) control system simulation program to study
HVAC control
system dynamics and interactions...
http://www.nist.gov/public_affairs/guide/enhanced_building.htm
and elswhere in the world
----------------------------------------
Public and Private Initiatives in Sustainable Building - 44
page report
www.cyberbites.com/dcat/Resources/ green_building_initiatives.pd
Here are 122 pages on modern trends in Control systems
www.cds.caltech.edu/~murray/cdspanel/
report/cdspanel-15aug02.pdf
___________
In conclusion
_____________
A system which would realize intelligent control of
heating the pool, heating and cooling house, and heating water
can be built, using a home computer, possibly an old slow computer
running linux OS and suitable control and simulation software.
It would use X10 protocol to cummunicate with sensors and
switches and possibly, over the internet, with weather forcasts.
It would require some custom programming (perl, tcl, or c)
or, perhaps, in time, it may use some adaptive control systems
which 'will measure and learn' the parameters of the system (thermal
inertia, time lag of heaters, weather patterns ..).
Study of the 'academic resources' listed above may lead to the source
of
such a program. When such program is located, one does not need the
set of equations and parameters for the house. AI programs such as
Neural Network can build suitable model 'on the fly' .
Such equations are available, if needed ( Request a clarification
on this, if you need it).
Simulation
====================================
search terms
computer model, heating, home simulation
Simulation of the whole plant ( house, solar panels, pool ..) is
highly recommended.
It not only allows one to test ideas and designs, but can be
eventually used as
a major part of the control system itself. Simulation of the whole
control
system and plant is an excellent tool for developing and testing the
design.
This site may have some data, but no computer program is apparent
http://www.eere.energy.gov/solarbuildings/publications.html
Energy Smart Pools Software
A tool to analyze ... tools for building analysis, HVAC equipment and
systems ...
www.es.wapa.gov/energy_solutions/
res_list.cfm?category=Software%20Tools&subcategory=%20General - 52k -
17 Sep 2003
This simulation was developed for greenhouses
It may be adaptable to a residence and pool
http://www.actahort.org/books/135/135_7.htm
This describes the use of VisSim modelling software
http://www.vissim.com/apps/dupont.htm
This is commercial software available for a free trial.
Here is a long list of simulation tools
http://216.239.59.104/search?q=cache:iVvv05ZmXrgJ:arch.hku.hk/research/BEER/best.htm+computer+simulation+model,+heating+home&hl=en&lr=lang_en&ie=UTF-8
and here is the home of DOE-2 and BLAST , two programs
mentione above.
One would have to spend time testing and evaluating the models they
developed
to the task being considered here.
Authors should be able to answer questions and provide guidance.
http://gundog.lbl.gov/dhomepage.html
Let me know what you think of these links.
If needed, please use the " request for clarfication" not only to
ask for further
explanations of points vovered here but to indicate directions for
further search which may be needed to complete the answer.
Please do clarify whether you are willing
to do any programing, wiring, oriwhether you would contract
that portion out, if a ready made solution is not found.
The Bay Area has plenty of resurces for custom programing
if that would be necessary.
hedgie |
