Nof S.Y. Springer Handbook of Automation

Подождите немного. Документ загружается.

The Smart Building 62.2 Application Examples 1085

a standard offering from one large controls vendor in

the USA.

Multiloop

Multiloop feedback control strategies are used in build-

ings, with the most common being cascaded control. In

these conﬁgurations, there is an inner loop that controls

a fast variable and an outer loop that controls the set-

point ofthe innerloop basedon thefeedback ofa slower

variable. An example is the control of variable-air-

volume (VAV) boxes for regulating room temperature,

as illustrated in Fig. 62.3. These boxes are supplied with

conditioned air from a central air-handling unit and

they can control the ﬂow of conditioned air to a room

by means of a damper. In the inner loop, the con-

trolled variable is the airﬂow rate entering the room.

In the outer loop, the controlled variable is the room

temperature. The outer loop tries to control the room

temperature to its setpoint by modulating the setpoint

for the airﬂow in the inner loop. In turn, the inner loop

tries to meet the airﬂow setpoint by modulating the

VAV box damper. Cascaded control allows a control

problem to be separated into two parts, one with slow

dynamics and the other with fast dynamics. Each type

of fast and slow disturbance is then handled by a sep-

arate controller, with improved performance compared

with having just one controller. A general guideline for

cascade control is that the inner loop should be at least

three times faster than the outer loop. Further guidelines

for implementation can be found in [62.14].

Multivariable control in the sense of having central-

ized controllers that are multiple-input multiple-output

(MIMO) is not typical in building applications. The

most common situation is to have multiple feedback

controllers that do not share information with each

other. For example, a large open-plan ofﬁce space might

have severalcontrolled temperaturezones thathave sep-

arate feedback controllers. An obvious problem with

this approach is when there are interactions between

zones due to effects such as interzonal airﬂow. Poor

performance in one zone can then propagate to other

zones if interactions are strong. Current practice in-

volves trying to minimize interactions by positioning

sensors away from each other and making sure that

control zones are separated.

There is generally signiﬁcant interaction amongst

building systems and in many cases operation could

be improved through application of multivariable con-

trol methods. Model-based approaches are one way of

handling multivariable systems and these methods have

been investigated and applied to buildings, e.g., [62.21].

Damper

motor

VAV

controller

Setpoint and

room temperature

Room

Flow station

Control signal

Thermostat

Fig. 62.3 Example cascaded control strategy

However, adoption of these methods by the build-

ing industry is impeded by the costs associated with

setup and commissioning and of handling the additional

complexity. Nevertheless, multivariable control is un-

derexploited in building applications and it appears to

be a promising direction for future research.

62.2.3 Energy Management Control

Strategies

Strategies for energymanagement are usually employed

at a higher level in the control hierarchy of a build-

ing. Many controls vendors offer proprietary algorithms

for energy management and these are normally pro-

grammed into so-called supervisory controllers.In

contrast to local controllers, which handle simple tasks

such as feedback loops, supervisory controllers have

more processing power and memory and consolidate

data from multiple nodes on a network. Supervisory

controllers perform functions that may affect the opera-

tion of several feedback controllers by adjusting things

such as setpoint and operation schedules. This section

reviews some of the most common energy manage-

ment control strategies that are found in buildings. The

overall idea behind the strategies is explained, but al-

gorithmic details are not provided because no standard

implementation exists and many different types of pro-

prietary implementations can be found in practice.

A variable that iscommonly usedin energy manage-

ment control strategies is the outside air temperature.

This variable is known to affect the thermal loads on

a building and its ultimate heating and cooling require-

ments. One example where the outside air temperature

is used as part of feedforward control strategy is the op-

timum start algorithm. This algorithm uses the outside

air temperature to determine when to turn on the heat-

ing and cooling systems in a buildingso that theinternal

temperatures will be at their respective setpoints by the

Part G 62.2

1086 Part G Infrastructure and Service Automation

time people occupy the building [62.22]. The algorithm

needs to predict the time required to heat or cool the

building from a known start temperature and outside

conditions and thus requires some kind of model of the

building’s thermal response. The complement of opti-

mum start is optimum stop, which determines a time

when systems can be turned off, allowing the capacity

of the building to hold conditions near their setpoints

until people leave.

Optimum start and stop strategies are often com-

bined with night-setback. Instead of just turning all

systems off during unoccupied periods, night setback

control maintains closed-loop control on space tem-

peratures but at different setpoints. During the heating

season, for example, temperatures inside the building

would be kept lower during the unoccupied period, but

only low enough that recovery time would be rapid

enough and losses of energy storedin building materials

minimized so that this stored energy can be carried over

to the next day to reduce overall energy use. Various

algorithms have been proposed for predicting the night-

setback setpoints and startup and shutdown times based

on models and optimization and the use of thermal stor-

age systems, e.g., [62.23,24]. Integrated optimum start

and stop with night-setback algorithms are available

from some control vendors, and tuning and setup re-

quirements depend on the particular algorithm that is

offered and the systems that are targeted for control.

Another higher-level control strategy that uses

feedforward is setpoint reset, where the setpoint of

a feedback loop is adjusted based on a feedforward vari-

able; for example, the setpoint of the main supply of

conditioned air to a building can be made a function

of the outside air temperature. The setpoint would be

adjusted to be higher when the outside air temperature

is low and made to be lower if the outside tempera-

ture is high. The setpoint of cold and hot water supplies

generated by building chiller and boiler plant are other

variables that are sometimes linked to reset strategies

based on outside air temperature. Most of these algo-

rithms adjust setpoints between two speciﬁed limits as

the outside air temperature varies between upper and

lower bounds. Setup and tuning of these reset strategies

usually requires speciﬁcation of the upper and lower

values for the setpoints and the outside air temperature.

Feedforward control can also be used to supplement

feedback control in situations where disturbances such

as changes in airﬂow and outside temperatures can be

measured in systems such as air-handling units [62.25].

The advantage of supplementingfeedback with feedfor-

ward is that the effect of disturbances can be mitigated

when they occurinstead ofhavingto wait fortheir effect

to be revealed in the feedback variable. It is unfortu-

nate that many disturbance signals in buildings are in

fact measured but not used in the control strategy. The

number of potentiallyuseful signalsis also increasing as

more types of systems become integrated in the BAS;

for example, signals from lighting occupancy sensors

could be used to improve the control of the HVAC sys-

tems byadjusting capacity based onanticipated changes

in loads.

Economizer control is a very common energy-

saving strategy, particularly in the USA, that is used to

reduce cooling loads by switching the air source to the

main air-handling systems between recirculated (with

a mix of minimum outside air) and 100% outside air.

An economizer strategyevaluates thetemperature or en-

thalpy of the two potential air sources and issues control

signals to dampers to provide an air supply to cool-

ing heat exchangers with the aim of reducing required

cooling capacity. This control method is usually part

of sequencing logic that is used to coordinate the op-

eration of heating, cooling, and recirculation systems.

The energy saving potential of economizer control has

been shown to be as high as 52% [62.26], but results

depend on climatic conditions. Both temperature and

enthalpy measurements are meant as proxies for the

eventual cooling load. Improved results could thus be

obtained by using more accurate methods for predicting

loads, e.g., by usingthermodynamic models of the cool-

ing plant. Today, control vendors normally only offer

a choice between temperature or enthalpy economizer

strategies. In most cases, enthalpy-based strategies will

yield more energy savings, but results can be inconsis-

tent because most humidity sensors, which are used to

calculate enthalpy, are notoriously inaccurate.

Peak load management is an aspect of energy man-

agement that has received increased attention in recent

years. This is mostly due to the fact that utility compa-

nies often tie energy tariffs to peak demand statistics,

thereby creating incentives to minimize peaks and ﬂat-

ten the load proﬁle. The strategy is commonly termed

demand limiting and involves shutting down the opera-

tion of certain plant items to keep demand below target

levels. The concept is illustrated in Fig.62.4,which

shows how building demand is kept below a target level

by shedding loads. The objective of this strategy is to

control peak demands while at the same time minimiz-

ing the disruption to control objectives such as comfort

conditions within the building [62.27]. Unresolved re-

search issues are how to set the targets and how to

determine the order and time over which equipment

Part G 62.2

The Smart Building 62.2 Application Examples 1087

Whole building load

Without demand

limiting

Load(s)

added

Load(s)

removed

With demand

limiting

Demand threshold

Time

Fig. 62.4 Demand limiting scenario

loads should be added and removed. Currently, the al-

gorithms available from control vendors for demand

limiting require the user to set targets based on expert

knowledge, which leads to very variable results from

these strategies.

The discussion so far has focused on energy man-

agement and control of HVAC systems. The reason for

this focus is that this group of systems usually repre-

sents the only example of where algorithms with more

sophistication than simple scheduling and feedback can

be found in contemporary buildings. Although more

system typesare being integrated into the BAS network,

the control strategies for these systems are normally

primitive and frequently just provide a means for cen-

tralized manual operation. However, the potential for

more sophisticated control and coordination of systems

such as lighting, ﬁre, and security has been recognized

for some time and examples of specialized buildings

with more sophisticated control strategies can be found.

Some of the ideas implementedin theseshowcasebuild-

ings are beginning to trickle down to the rest of the

building stock but the rate of adoption is slow, mostly

because of cost constraints. Lighting systems are one

example of where the use of more sophisticated con-

trol can overcome cost constraints by reducing energy

use. Traditionally, lights have been controlled manually

as a form of open-loop control. However, studies have

shown that the implementation of automated feedback

signals such as occupancy measurements and outside

light sensors can be very effective for control and en-

ergy reduction [62.28]. More advanced control is also

starting to appear, making use of additional controllable

elements such as shading devices that can be used to

regulate the inﬂow of natural light into a space [62.29].

This section has only provided a brief overview of

a sample of energy management strategies in build-

ings. Further examples related to HVAC can be found

in [62.30]. The demand for these kinds of strategies

is growing in response to higher energy prices and in-

creased environmental concerns, and the computing and

networkinginfrastructure needed toperform thiskind of

plant-wide control is becoming more widely availably.

Most energy management strategies are implemented

as supervisory logic on top of lower feedback con-

trol loops. In controls terminology the combination of

this kind of supervisory logic with low-level feedback

control is known as hybrid control. Design of hybrid

control strategies inbuildings is becoming more formal-

ized due to the availabilityof new software design tools;

for example, the lower-level control strategies might be

designed using block diagrams and transfer functions

and the supervisory logic designed using state-machine

logic diagrams. Modern software programs also allow

control logic to be simulated before being implemented,

which greatly improves reliability and lowers develop-

ment costs.

62.2.4 Performance Monitoring and Alarms

Control and operationalautomation is themost common

application of intelligence in buildings. However, recent

years have seen a growing interest in intelligent moni-

toring and analysis of operational performance [62.31].

This has paralleled similar research and application in

other industries such as aerospace, chemical process-

ing, and power generation. The most pressing need for

monitoring and analysis technologies is to detect faults

that could jeopardize performance and, more impor-

tantly, safety. The concept of generating an alarm when

a critical variable is outside of acceptable bounds is the

simplest implementation and is employed quite widely

in modern buildings. Although the idea is simple, set-

ting the alarm limits requires intelligence and in some

cases may need to be periodically adjusted to cater for

changing conditions. A detailed discussion on safety

warnings in automation is provided in Chap.39.

Recent years have seen a demand for more sophis-

ticated performance monitoring and analysis, beyond

simple alarming [62.32,33]. The motivation for this de-

rives from a recognition that control and automation

of system operation frequently fall short of expecta-

tions. Poor performance can be caused by many things,

ranging from faults and deteriorations in the plant,

badly tuned controllers, wrongly implemented control

logic, to sensor or actuator malfunction. There is also

a drive to increase the automation of analysis and oper-

ational oversight tasks that were previously carried out

manually, not only to reduce costs, but also to help op-

erators deal with the overwhelming amount of data now

Part G 62.2

1088 Part G Infrastructure and Service Automation

Comparison/

analysis

,..., E

Expectations

,..., p

Performance

variables

Transformation

Results

Data

Fig. 62.5 Example performance assessment methodology

available to them through modern building automation

systems.

The associated problems of performance analysis

and fault detection and diagnosis primarily involve

transforming data available from sensor measurements

and control actions into variables related toperformance

that can be compared with expectations, as illustrated

in Fig.62.5. An example of a performance variable

is the coefﬁcient of performance (COP) of a chiller,

which is calculated from several temperature and power

measurements. Fault detection involves comparing per-

formance variables with expectations and making

a binary decision as to whether a fault exists or not.

Fault detection is therefore very similar to simple alarm-

ing except that the variables being monitored are not

raw measurements, but derived from them using mod-

els, statistics, expert rules or some other transformation

method. Fault diagnosis is more complicated and can

be broken down into locating the source of the problem,

matching symptoms to a cause, and estimating the mag-

nitude of the fault. In a building, a hierarchical approach

might start with the detection of higher than normal en-

ergyuse, get narrowed down to oneair-handlingsystem,

and subsequently to a valve on a heat exchanger. Fi-

nally the fault might be diagnosed as a valve leakage

and estimated to be 50% of maximum ﬂow.

The problem of performance monitoring in build-

ings is also suitable for the application of single-input

single-output (SISO) loop monitoring techniques de-

veloped for other industries. Measuring the variance

of controlled variables and comparing with theoreti-

cal benchmarks for example is one method that has

been successfully deployed in many industries [62.34].

However, variance of control variables is normally less

important in buildings than in other manufacturing-

type industries where product quality and proﬁt is

directly correlated with control variable variance. One

of the biggest problems in buildings lies in detect-

ing problems with the plant such as leaking valves,

stuck dampers,sensor failures, andother moreprevalent

malfunctions [62.31].

62.3 Emerging Trends

As mentioned at the start of this chapter, buildings can

be viewed as large processes with many interacting and

diverse systems operating together to achieve various

objectives. In this way buildings are broadly similar to

other applications such as chemical processing, power

generation or oil reﬁneries. A building has numerous

Manipulated variables Measured variables

Algorithms

Actuators

Plant Environment

Sensors

Fig. 62.6 Linking the measurements

to the manipulated variables

manipulated variables and measured signals like these

other systems. The challenge lies in how to connect

the variables together and what algorithms and logic to

deploy between these linkages. Figure 62.6 illustrates

the concept of how algorithms and logic routines link

manipulated and measured variables.

Part G 62.3

The Smart Building 62.3 Emerging Trends 1089

One trend that is still emerging in some sectors

of the buildings industry but is already mature in

other sectors is the deployment of IT infrastructure to

provide a backbone that allows all sensors actuators

and other devices to be visible on a single network.

This is part of a more general trend toward the con-

vergence of the building automation and information

technology systems in a building. Having an IT infra-

structure in place that links disparate building devices

facilitates the development of advanced control and op-

erational management algorithms that take advantage

of the plant-wide data. Energy management strategies

that were mentioned in Sect.62.2.3 are one example

of algorithms that combine information from different

subsystems to attain an overall objective. Application

of these types of building-wide optimization is likely to

increase in the future with the demand driven by rising

energy costs and increasing environmental concerns.

IT infrastructure and networking is constantly

evolving, making it easier to connect and redistribute

devices; for example, wireless networking is becom-

ing more popular in buildings because it reduces wiring

costs and makes it easier to reconﬁgure spaces and

move sensors from one location to another. The gen-

eral trend is also for a greater diversity of devices to be

connected to the network and for each device to contain

some level of embedded intelligence. These so-called

smart devices are part of a trend toward distributed

computing. Distribution of computing functions across

devices makes a system less prone to catastrophic fail-

ure and allows problems to be broken up into smaller

pieces and solved using many low-cost devices rather

than one high-cost computing engine.

The developmentof automationalgorithms andcon-

trol techniques has lagged that of hardware and IT

infrastructure. Many opportunities therefore now exist

for making better use of the information available from

a building automation network to improve operation and

control and address higher-level objectives such as min-

imizing energy use. One particularly underdeveloped

topic is that of coordinating the operation of differ-

ent building systems; for example, taking into account

the interactions between lighting and HVAC could po-

tentially yield large energy savings for many types of

buildings [62.35]. Another example is food retail build-

ings that use refrigeration systems to keep products

cool. Refrigeration systems generate heat and affect

the indoor environment in these buildings but their

operation is not usually coordinated with the HVAC

plant. Combining information from security and ac-

cess control systems to predict the number of people in

a space for improved climate control is yet another ex-

ample, mentioned earlier, of how information from one

subsystem could be used to enhance the operation of

another.

Continuing with the theme of IT infrastructure

providing new opportunities for higher level-plant

coordination, recent years have also seen the deploy-

ment of multibuilding control and operation manage-

ment [62.27]. The concept here is to tie together several

buildings and have operational oversight and alarm

management handled in one location rather than in each

individual building. This allows for centralized data

processing and the possibility for a smaller group of

highly qualiﬁed operators to spread their expertise over

multiple buildings. The approach also opens the way

for peer-basedbenchmarking andperformance monitor-

ing that can be particularly advantageous when many

buildings are of the same type [62.36].

The discussion so far in this section has identiﬁed

some potential opportunities for control and operational

management that are the result of having more in-

formation from multiple diverse systems available on

a network. Capitalizing on these opportunities often

requires combining ideas and methods from different

mathematical disciplines. This has been an emerging

trend in recent years with a common example being

the combination of statistics and control theory such

as when employed in statistical process control [62.37].

Other examples are the use of economic ideas as a way

to instigate distributed optimization. The availability of

real-time energy prices in certain geographical areas is

an example of using economics to encouragedistributed

optimization ofload proﬁleswith the aim of eveningout

the loads at the power plant [62.38].

The idea of using market-based theories is also be-

ginning to be seen as a viable strategy for optimizing

the operation of systems within a building. The way in

which building systems are designed and implemented

is inherently distributed, with small amounts of pro-

cessing power usually available locally at each device

rather than concentrated in a central location. Further-

more, the number of variables in a building and the

complexity of interactions and nonlinear behaviors can

make a centralized approach to optimization unviable.

The metaphor of each device being an agent working

to optimize its own objective function according to con-

straints and general guidelines is an emerging ﬁeld of

research with the aim being that overall building per-

formance would reach some optimum behavior without

the need for centralized optimization. This idea is be-

ginning to be explored for building applications [62.39]

Part G 62.3

1090 Part G Infrastructure and Service Automation

0 5 10 15 20 25

EWMA of |flow error|

Maximum flow rate

Possible

problem

VAV box number

1.2

0.8

0.6

0.4

0.2

Fig. 62.7 Normalized EWMA statistics for VAV box con-

trol performance

and has already been applied to problems in other

applications [62.40].

The expansion of control methods and plant oper-

ational management ideas that has occurred in other

applications is highly relevant to buildings and many of

the most promising ideas are beginning to be explored

and adopted. The general subject of plant perfor-

mance assessment and auditing is one example where

advances from other applications are being adopted.

Some control vendors now offer automatic trending and

benchmarking of performance indices that are not just

raw measurements, but quantities derived from physics

and/or statistics.The methods of analyzing these perfor-

mance indices and ways in which they can be presented

to a user are emerging areas of research and application.

The goal of this work is to be able to quickly detect

problems and narrow down the cause so that down-

times are reduced, performance is more consistent, and

maintenance can be more proactive and targeted. One

popular approach is to group together operational statis-

tics from multiple plant items that are of similar type

and lookfor outliers amongst the set asa wayto identify

faults. Figure 62.7 illustrates this approach based on ac-

tual exponentially weighted moving averages (EWMA)

of the setpoint error signal for VAV boxes in a large

building [62.41].

An extension of the concepts of detecting and di-

agnosing problems is to make the control system able

to adjust automatically so that the building can become

fault-tolerant. Buildings usually have many different

types of systems that can be used to compensate for

each other; for example, a problem identiﬁed with

cooling capacity that is affecting the temperature of

conditioned air could be compensated for by operat-

ing the fans at higher loads to increase airﬂow. This

redundancy is already (inadvertently) used to create

fault tolerance in buildings by having multiple closed-

loop control strategies that sometimes compete with

each other. This creates problems for more conventional

methods of fault detection that only check whether

setpoints are being maintained because the effects of

a fault are masked by the competing closed-loop con-

trollers. Hence, there is increasing awareness that fault

detection and diagnosis methods need to share informa-

tion with the control strategy and be combined for the

most effective solution [62.42].

The buildings industry is notoriously fragmented

with multiple parties involved in the various lifecycle

tasks of, among others: design, construction, occupa-

tion, maintenance, and renovation. An emerging area is

the application of information management to a build-

ing lifecycle. Recent years have seen concerted efforts

to standardize various aspects of a building process,

ranging from standard data models for building geom-

etry and plant description [62.43] to communication

protocols for automation system networks [62.44]. The

general goal of these standardization efforts is to im-

prove the efﬁciency of passing information and also

reduce the costs and risk of developing and market-

ing new products such as software. The proliferation

of web-based commerce is testament to how standard

(and stable) infrastructure can be an effective stimulant

to innovation and technological progress.

62.4 Open Challenges

Although buildings can be considered similar to any

other large-scale process, there are some unique is-

sues that can hamper efforts to make buildings smart.

Primary systemic issues in the building industry are

that it is low cost, fragmented, and risk averse. The

way in which buildings are ﬁnanced is geared to-

ward minimization of capital costs. Most contracts

are awarded on a lowest-cost basis and this means

that the operational costs of a building and other

lifecycle costs are deemphasized. The result is that

Part G 62.4

The Smart Building 62.4 Open Challenges 1091

operational performance of the building is frequently

compromised through poor design and poor-quality in-

stallation and maintenance. Several studies have shown

that proper commissioning of buildings can lead to

large energy savings and improvements in system per-

formance [62.45]. Cost pressures also lead to the

installation of low-quality sensors and actuators that

detrimentally affect control performance. A related

problem is when too few sensors are installed, causing

measurements to be inaccurate and control performance

and energy use to be affected.

At the control hardware level, cost constraints lead

to minimal memory and processing power in local con-

trollers and also low-resolution analog-to-digital and

digital-to-analog converters. Eight-bit converters are

still commonplace and the level of signal quantization

can be severe enough to cause oscillations in feed-

back loops. Quantizationcan alsocorrupt signals,which

makes later analysis and diagnosis of problems more

difﬁcult. Another source of quantization is logic that

is included on a network so that data are only sent

when there is a change of value beyond a certain

threshold. This strategy provides data compression and

network trafﬁc minimization but it severely affects the

signals being ﬁltered and poses problems when trying

to analyze data for control performance assessment and

diagnostics [62.46].

The level of education and training tends to be low

for building operators, also because of cost constraints.

This means that there is often difﬁculty in understand-

ing the processes in a building, their interactions, and

the role of the control system. It is common therefore

for operators to quickly shut-off control logic that they

do not properly understand. The situation then arises of

having many loops in a building left in override mode,

leading to suboptimal and effectively open-loop, or no,

control of the building systems. The primary objective

of most operators is also to keep the occupants com-

fortable and respond to hot and cold complaints. The

energy efﬁciency of the building is usually secondary

and draws little attention. The lack of consideration for

energy use is due to the still relatively cheap cost of en-

ergy and also the lack of performance indicators that

could help identify energy problems.

The discussion so far has centered on barriers,

both industry-systemic and technical, that impede the

development, application, and deployment of new

smart-building algorithms and technologies. However,

many of these barriers can be overcome by adapting

smart control and operational management methods to

the unique aspects of the buildings industry; for exam-

ple, the problem of too few sensors can be addressed by

using models to predict measurements to create virtual

sensors based on analytical redundancy. Signal pro-

cessing methods can be used to reconstruct quantized

data, and robust control theory can be used to minimize

the effect of slow sampling and potential jitter. Build-

ings can beneﬁt from the application of methods and

algorithms developed in other diverse ﬁelds but these

methods have to be adapted to the speciﬁc issues of the

buildings.

Another open challenge is to integrate the di-

verse array of systems in buildings into a common

information-sharing framework. Although this has al-

ready started to happen with the advent of open

protocols for network communication, the full potential

is not yet realized. One reason for this is that build-

ings contain a very diverse group of systems that are

made by many different manufacturers, all with their

own embedded controls and electronics. The commodi-

tization of certain building systems such as packaged

cooling units has brought down costs, but these sys-

tems rarely have external communication interfaces to

allow interconnectivity with other systems because of

the extra costs this would incur. Although interconnec-

tion of building devices is a key to unlocking untapped

energy and operational efﬁciencies, this cannot happen

without algorithms and control methods that can take

advantage of the newly available information. There is

therefore a dilemma because both aspects are needed,

and business will be reluctant to invest in develop-

ment of just one aspect without the other being already

available. Possible future avenues that could alleviate

some of the barriers to connectivity are solutions such

as power line networking, utilization of an existing IT

backbone, and general lower-cost solutions for plug-

and-play networking.

The issue of energy efﬁciency has again come

to the fore in recent years, and this has led to new

legislation and various certiﬁcation programs being

established in several countries around the world to

encourage the design and construction of so-called

green buildings. Even if ﬁnancial incentives are lack-

ing, increased concern over the environmental impacts

of wasteful energy use within populations is exert-

ing pressure on the construction industry to portray

a greener and more energy-efﬁcient image. The most

visible sign of a response to these pressures is in the de-

sign and construction of modern buildings. There are

several examples of new buildings around the world

that make such efﬁcient use of sunlight and wind

through smart design that they can eliminate the need

Part G 62.4

1092 Part G Infrastructure and Service Automation

for mechanical heating/cooling and ventilation and al-

most eliminate the need for artiﬁcial lighting when

sunlight is available. The manufacturers of electrical

and mechanical systems are also responding to energy

concerns by producing more efﬁcient equipment that

is often certiﬁed by government or other third-party

organizations. However, the challenge isfor the automa-

tion systems and especially the control algorithms to

keep pace with the rapid evolution of buildings and in-

herent equipment. Failure to keep pace will result in

buildings operating well below their intended efﬁciency

levels.

62.5 Conclusions

The operation of many aspects of a building is now

automated, ranging from temperature control to ﬁre

and security. The general trend is for the operation

of more systems and appliances to be automated and

for these to be connected to a common network. The

availabilityof commonnetworking infrastructure isalso

stimulating the demand for more advanced control and

operational management methods that take into account

system interactions and facilitate the optimization of

building-wide criteria such as energy use. However,

the adoption of IT infrastructure has in many ways

outpaced the development of algorithms and automa-

tion methods that can capitalize on these advances.

There is also the problem of operators not being able

to cope with the abundance of data now available on

BAS networks. Again, this problem is exacerbated by

the lack of algorithms for processing and reducing the

data into more manageable statistics or performance

reports.

The complex and bespoke nature of building sys-

tems makes it difﬁcult to develop and apply generic

algorithms and, on the other hand, it is too costly to

tailor algorithms to every building system. This issue

is difﬁcult to resolve but is being alleviated by stan-

dardization and commoditization of building systems.

New model-free algorithmic methods for control and

optimization have the potential of being able to adapt

to different system types without requiring signiﬁcant

engineering effort or tuning. In combination, these de-

velopments may help overcome some of the barriers to

the adoption of new technology for improved energy

management and control in buildings.

References

62.1 IEA: International Energy Outlook 2007 (United

States Department of Energy, Washington 2007),

retrieved on 2007-06-06

62.2 H.M. Newman: Direct Digital Control of Building

Systems: Theory and Practice (Wiley, New York 1994)

62.3 ASHRAE: HVAC Systems and Equipment (Amer-

ican Society of Heating, Ventilating, and Air-

Conditioning Engineers, Atlanta 2004)

62.4 WHO: Health Aspects of Plumbing (World Health Or-

ganization and World Plumbing Council, Geneva

2006)

62.5 G. Augenbroe: Trends in building simulation,

Build. Environ. 37(8-9), 891–902 (2002)

62.6 Y. Zhang, J.A. Wright, V.I. Hanby: Energy aspects of

HVAC system conﬁgurations – problem deﬁnition

and test cases, HVAC&R Research 12(3C), 871–888

(2006)

62.7 J. Sinopoli: Smart Buildings (Spicewood Publishing,

Elk Grove Village 2006)

62.8 I. Shaw: The Oxford History of Ancient Egypt (Oxford

Univ. Press, Oxford 2003)

62.9 J.W. Humphrey, J.P. Oleson, A.N. Sherwood: Greek

and Roman Technology: A Sourcebook (Routledge,

London 1997)

62.10 W.H. Carrier: Modern Air-Conditioning, Heating

and Ventilating, 2nd edn. (Pitman Publishing, New

York 1950)

62.11 O. Mayr: The Origins of Feedback Control (Cam-

bridge MIT Press, Cambridge 1970)

62.12 L.E. Harris: The Two Netherlanders, Humphrey

Bradley and Cornelis Drebbel (Cambridge Univ.

Press, Cambridge 1961)

62.13 W.S. Johnson: Electric Tele-Thermoscope, Patent

281884 (1883)

62.14 K.Åström,T.Hägglund:PID Controllers: Theory, De-

sign and Tuning, 2nd edn. (Instrument Society of

America, Research Triangle Park 1995)

62.15 T.I. Salsbury: A new pulse modulation adaptive

controller (PMAC) applied to HVAC systems, Control

Eng. Pract. 10(12), 1357–1370 (2002)

62.16 T. Marlin: Process Control, 2nd edn. (McGraw–Hill

Education, New York 2000)

Part G 62

The Smart Building References 1093

62.17 A.L. Dexter, R.G. Hayes: Self-tuning charge con-

trol scheme for domestic stored-energy heating

systems, IEE Proc. D: Control Theory Appl. 128(6),

292–300 (1981)

62.18 X.-C. Xi, A.-N. Poo, S.-K. Chou: Support vector re-

gression model predictive control on a HVAC plant,

Control Eng. Pract. 15(8), 897–908 (2007)

62.19 S.J. Hepworth, A.L. Dexter: Adaptive neural control

with stable learning, Math. Comput. Simul. 41(1-2),

39–51 (1996)

62.20 J.E. Seem: A new pattern recognition adaptive

controller with application to HVAC systems, Au-

tomatica 34(8), 969–982 (1998)

62.21 X.-D. He, S. Liu, H.H. Asada: Modeling of vapor

compression cycles for multivariable feedback con-

trol of HVAC systems, J. Dyn. Syst. Measur. Control

Trans. ASME 119 (2) 119(2), 183–191 (1997)

62.22 A.L. Dexter: Self-tuning optimum start control of

heating plant, Automatica 17(3), 483–492 (1981)

62.23 J.M. House, T.F. Smith: System approach to opti-

mal control for HVAC and building systems, ASHRAE

Transactions 101(2), 647–660 (1995)

62.24 F.B. Morris, J.E. Braun, S.J. Treado: Experimen-

tal and simulated performance of optimal control

of building thermal storage, ASHRAE Transactions

100(1), 402–414 (1994)

62.25 T.I. Salsbury: A temperature controller for VAV

air-handling units based on simpliﬁed physical

models, HVAC&R Research 4(3), 265–279 (1998)

62.26 J.D. Spitler, D.C. Hittle, D.L. Johnson, C.O. Peder-

sen: A comparative study of the performance of

temperature-based and enthalpy-based economy

cycles, ASHRAE Transactions 93(2), 13–22 (1989)

62.27 T.A. Reddy, J.K. Lukes, L.K. Norford, L.G. Spielvogel,

L. Norford: Beneﬁts of multi-building electric load

aggregation: actual and simulation case studies,

ASHRAE Transactions 110(2), 130–144 (2004)

62.28 B. Von Neida, D. Maniccia, A. Tweed: An analysis of

the energy and cost savings potential of occupancy

sensors for commercial lighting systems, J. Illum.

Eng. Soc. 30(2), 111–125 (2001)

62.29 A. Guillemin, N. Morel: Innovative lighting con-

troller integrated in a self-adaptive building

control system, Energy Build. 33(5), 477–487

(2001)

62.30 ASHRAE: HVAC Applications (American Society of

Heating, Ventilating, and Air-Conditioning Engi-

neers, Atlanta 2007)

62.31 J. Hyvarinen, S. Karki: Final Report Vol 1: Build-

ing Optimization and Fault Diagnosis Source Book

(Technical Research Centre of Finland, Espoo 1996)

62.32 S. Katipamula, M.R. Brambley: methods for fault

detection, diagnostics, and prognostics for build-

ing systems - a review, Part II, HVAC&R Research

11(2), 169–187 (2005)

62.33 S. Katipamula, M.R. Brambley: Methods for fault

detection, diagnostics, and prognostics for build-

ing systems - a review, Part I, HVAC&R Research

(1), 3–25 (2005)

62.34 T.J. Harris: Assessment of control loop perfor-

mance, Can. J. Chem. Eng. 67, 856–861 (1989)

62.35 O. Sezgen, J.G. Koomey: Interactions between

lighting and space conditioning energy use in

US commercial buildings, Energy 25(8), 793–805

(2000)

62.36 K.L. Gillespie Jr., P. Haves, R.J. Hitchcock, J. De-

ringer, K.L. Kinney: Performance monitoring in

commercial and institutional buildings, HPAC En-

gineering 78(12), 39–45 (2006)

62.37 J. Schein, J.M. House: Application of control charts

for detecting faults in variable-air-volume boxes,

ASHRAE Transactions 109(2), 671–682 (2003)

62.38 D.S.Watson,M.A.Piette,O.Sezgen,N.Motegi:Au-

tomated demand response, HPAC Engineering 76,

20–29 (2004)

62.39 P. Davidsson, M. Boman: Distributed monitor-

ing and control of ofﬁce buildings by embedded

agents, Inf. Sci. 171,293–307(2005)

62.40 N.R. Jennings, S. Bussmann: Agent-based control

systems, IEEE Control Syst. Mag. 23(3), 61–73 (2003)

62.41 J.E. Seem, J.M. House, R.H. Monroe: On-line mon-

itoring and fault detection, ASHRAE Journal 41(7),

21–26 (1999)

62.42 J.E. Seem, J.M. House: Integrated control and

fault detection of air-handling units, Proc. IFAC

Conf. Energy Sav. Control Plants Build. (Bulgaria

2006)

62.43 T. Froese, M. Fischer, F. Grobler, J. Ritzenthaler,

K. Yu, S. Sutherland, S. Staub, J. Kunz: Indus-

try foundation classes for project management -

a trial implementation, Electron. J. Inf. Technol.

Constr. 4, 17–36 (1999)

62.44 S.T. Bushby: BACnet: a standard communication

infrastructure for intelligent buildings, Autom.

Constr. 6(5-6), 529–540 (1997)

62.45 E.Mills,N.Bourassa,M.A.Piette,H.Friedman,T.

Haasl, T. Powell, D. Claridge: The cost-effectiveness

of commissioning, HPAC Engineering 77(10), 20–25

(2005)

62.46 N.F. Thornhill, M. Oettinger, P. Fedenczuk:

Reﬁnery-wide control loop performance assess-

ment, J. Process Control 9, 109–124 (1999)

Part G 62

“This page left intentionally blank.”