Plug and Play The Challenge for Diagnostics and Commissioning

IBS Plug and Play
Tools are the Challenge
in Creating IBS Products
• Many separate modeling tools exist
now
– Air flow modeling
– Thermal loads modeling
– Control & systems modeling
• Next generation of integrated
products require an integrated
research and development tool
The National Institute for Science and Technology
Building Fire Research Laboratory (NIST BFRL) has
developed an excellent air dispersion modeling system
called CONTAM for simulating fires in buildings.
However this system does not take into consideration
the effects of heating and cooling system and the
control system on the air flow in the building.
The US Department of Energy has created DOE-2 to
model energy usage in buildings. However, the tool is
meant to estimate hourly loads and does not consider
the detailed functioning of controllers and systems
such as the HVAC system. In addition it does not
produce data for faulted conditions.
NIST created HVACSIM+ & U. Wisonconsin created
TRNSYS to model the detailed functioning of HVAC
systems including faulted operation. However, these
systems do not include variable air flow volumes or
reverse flows and are not easily integrated with
control system models.
Control system vendors have several proprietary
controls simulation systems (SIMULINK is built with
MatLab). However, these systems are difficult to
integrated with air flow, thermal, control, systems, and
the new Plug & Play infrastructure we are developing.
Air Flow Modeling
(e.g., CONTAM)
Loads Modeling
(e.g., DOE-2)
Systems Modeling
(e.g., HVACSIM+)
Controls Modeling
(e.g., SIMULINK)
Integrated Modeling
(e.g., IBS Plug & Play)
The IBS Plug & Play
Modeling System
• Provides an integrated building modeling
environment to model interactions between
various systems in building in great detail.
• Can model faulty behavior and its impacts.
• Expandable from air, thermal, and control
system to include other elements such as
lights, equipment, and processes.
• Provide interfaces to users such as
operators, vendors, and service providers.
Plug and Plug Modeling Window
The Plug & Play system integrates many building
modeling tools. This allows IBS product developers to
build Plug & Play-enabled demonstrations, prototypes,
and products.
IBS product developers need data that represent both
faulted and un-faulted building operations. Software
tools based on learning systems such as neural
networks and genetic algorithms are an effective
means to quickly developing building system
diagnosticians, but they must have access to a very
large volume of reliable examples of both good and
bad behavior before they can “learn” to detect
problems in building systems.
The current Plug & Play system can only handle airside systems and controllers. The infrastructure is
open. Before a complete suite of IBS product can be
developed to support the Plug & Play system, it must
be extended to emulate and control other systems such
as hot and cold water distribution, chiller/boiler plants,
steam, electricity, lights, plugs, etc.
Plug & Play systems will have different user interfaces
for different users. For example, building operators
will have an operations console from which they might
override control signals. Control vendors will have a
controls design console with which they can specify
the control scheme and set points for the devices which
are found on the control network. Service providers
will have specialized consoles (such as the
commissioning console picture below) with which they
will provide services to building owners.
Controls Vendor Interface
Service Provider Interface
Simulation methodology
Plug & Play emulation handles variable sub-second
simulation time-steps for individual components (intracomponent) and 1-second simulation time-steps for
system level interactions (inter-component)
Windows-based network implementation
The Plug and Play system is implement in an extended
computational environment implemented on a minimum
of one Windows 95/98/NT-based computer. More
computers may be used if desired to implement
commissioning system, diagnostic systems, etc.
Systems Operator Interface
Implementation of faulted behavior
Development extensibility and compatibility
Faulted behavior can be introduced into any component
modeled in the system. For example a sensor can fail to
return a signal at all, or it can return an incorrect signal.
Or flow directions can reverse from the expected. The
effect of these faults will affect the other components in
the system appropriately.
The Plug & Play system is created using Microsoft
Visual C++ and SWI-Prolog. Communications
protocols are implemented using TCP/IP and Windows
sockets. The Plug & Play system is compatible with the
Whole-Building Diagnostician’s BASlink protocol. The
Plug & Play system appears on networks as a normal
building control system.
IBS Plug & Play
Product Application Areas
• Enable development and detailed evaluation
of energy savings of IBS products that rely
on benefits of integrating systems.
• Advance the state of the art in control
systems
– Enable next generation control in which
commissioning and diagnostic are
integrated.
– Automated control code generation from
heating/cooling system design specification.
– Generate compiled control code to reduce
size and cost of controllers.
• Create training data required to quickly
implement effective neural-net and genetic
algorithm-based diagnosticians
• Enable development of advanced and
automated commissioning procedures and
tools.
Today we cannot quantify the positive effects of IBS
technologies on the energy performance, occupant
health, and environmental impact of buildings. Plug
and Play development tools will permit the evaluation
of the economic and health benefits of IBS
technologies.
The most advanced control code development system
envisioning so far relies heavily on the existence of a
Plug and Play systems. Soon building designers will
specify what equipment they want and how they’d like
it work. Based on this design information, a suite of
Plug and Play tools will automatically generate the
control code, automated commissioning test
procedures, and diagnosticians without human
intervention. The result software would be deployed
more quickly, more reliably, and at lower cost than
present-day control systems.
Today’s control code relies on interpreters instead of
compilers. Compilers produce much more compact
and reliable code than code written for interpreters.
Smaller and more reliable code fits on smaller, lowercost controllers.
Diagnostician can be created manually but only a great
cost. A few diagnosticians have been created using
automated methods but they have generally either been
less reliable or unable to explain their decisions. The
available of more and better training data will enable
the development of better neural network-based
diagnostician and the development of the a new type of
genetic algorithm-based diagnostician capable of
providing explanations and cause isolation.