IMPROVING ENERGY UTILIZATION:

A Study of the Maxwell Dworkin Laboratory

Division of Engineering and Applied Sciences
Harvard University, Cambridge, Massachusetts

John An Aruna Balakrishnan

Andrew Carlson Jared Causer
Rohan Gulrajani John Jones
Roy Kaiser Mary Krause
Timothy Mariano Laura Thomson
Aaron Wilson

ENGINEERING SCIENCES 96: DESIGN SEMINAR SPRING 2002

ii
ACKNOWLEDGEMENT
The ES 96 Class would like to express our gratitude to the many people who so generously assisted
us in our investigation. In particular, we would like to thank:

• Prof. Michelle Addington

• Peter Arvidson
• Prof. Michael Brandstein
• Prof. Roger Brockett
• Susyrati Bunanta
• Frank DeCosta
• Jeff Deyette
• Armond Diaz
• Scott Gaines
• Prof. Barbara Grosz
• Jean Humber
• Ed Jackson
• Jonathan Kanda
• Greg Kousidis
• Xuan Liang
• Prof. David Parkes
• Jay Phillips
• David Richards
• Stephen Robichaud
• Dr. Joy Sircar
• Joe Ustinowich
• Prof. Gu-Yeon Wei
• Prof. Woodward Yang
• Rich Zitola

and, of course, our dedicated Coaching Staff,

• Prof. Frederick H. Abernathy
• Prof. Al Pandiscio
• Aaron Dollar
for sharing their knowledge, dedicating their time, and being so patient throughout the semester.

iii
iv
TABLE OF CONTENTS

Foreward . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . i

Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii

Table of Contents . . . . . .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . v

1.0 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

2.0 Energy & Environmental Indicators . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1 Energy, Economy, and Environment . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Environmental Issues and Opportunities . . . . . . . . . . . . . . . . . . . . . 7
2.1.2 Indicators of Energy and Environmental Efficiency . . . . . . . . . . . . . . . 8
2.2 Acres of Forest as an Environmental Indicator . . . . . . . . . . . . . . . . . . . . 9
2.2.1 Electricity to Acres of Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 Chilled Water to Acres of Forest . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.3 Steam to Acres of Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.0 Environmental Impacts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.1 Environmental Impacts of Maxwell Dworkin . . . . . . . . . . . . . . . . . . . 15
3.2 Energy Use in Maxwell Dworkin and Harvard Buildings . . . . . . . . . . . . . 17
3.3 Energy Consumption in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . . . 18

4.0 Building Energy Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

4.1 Building Scale Energy Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

5.0 Intelligent Control of Building Systems . . . . . . . . . . . . . . . . . . . . . . 29

5.1 APOGEE: The Intelligent Control System . . . . . . . . . . . . . . . . . . . . . . 31
5.1.1 System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
5.1.2 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
5.1.3 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.1.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
5.2 Air Ventilation and Circulation Systems Controlled by APOGEE . . . . . . . . . . 34
5.2.1 Air Handling Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.2.2 Fan Coil Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

6.0 Office Energy Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

6.1 Adaptable Temperature Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.1.1 The Status Quo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
6.2 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
6.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
6.2.2 Heat Capacity of a Room . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
6.2.3 Temperature Invariant Heat Sources . . . . . . . . . . . . . . . . . . . . . . 45
6.2.4 Heat Exchanges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
6.2.5 Active Temperature Regulation . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.2.6 Putting it All Together . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
6.2.7 Predicting the Status Quo . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
6.3 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
6.3.1 Application of the Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

v
 6.3.2 Optimizing the Office Controls . . . . . . . . . . . . . . . . . . . . . . . . . 53
6.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

7.0 Cooling in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
7.1.1 Chilled Water System in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . 61
7.2 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
7.3 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
7.3.1 Daily Use of Chilled Water . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.3.2 Weekly Use of Chilled Water . . . . . . . . . . . . . . . . . . . . . . . . . . 66
7.3.3 Monthly Use of Chilled Water . . . . . . . . . . . . . . . . . . . . . . . . . . 67
7.3.4 24-Hour Cooling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
7.4 Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
7.4.1 More Efficient Operation of the APOGEE Control System . . . . . . . . . . . . 70
7.4.2 Elimination of Weekend Cooling . . . . . . . . . . . . . . . . . . . . . . . . 71
7.4.3 Summer Night-Time Air Purges . . . . . . . . . . . . . . . . . . . . . . . . . 75
7.4.4 Ethylene Glycol Heat Exchangers . . . . . . . . . . . . . . . . . . . . . . . . 79
7.4.5 Steam to Hot Water Heat Exchanger . . . . . . . . . . . . . . . . . . . . . . 80
7.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81

8.0 Ventilation and Indoor Air Quality . . . . . . . . . . . . . . . . . . . . . . . . . 83

8.1 Introduction to Indoor Air Quality . . . . . . . . . . . . . . . . . . . . . . . . . 85
8.2 Determining Indoor Air Quality by Carbon Dioxide Levels . . . . . . . . . . . . 86
8.2.1 CO2 in Human Respiration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
8.2.2 Building Ventilation Requirements . . . . . . . . . . . . . . . . . . . . . . . 86
8.2.3 Predictive CO2 Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
8.2.4 Conclusion and Recommendations . . . . . . . . . . . . . . . . . . . . . . . 90
8.3 Indoor Air Quality via Massachusetts Building Code . . . . . . . . . . . . . . . 90

9.0 Air Handling Units in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . . . 93

9.1 Air Handling Units in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . . . . 95
9.2 Method of Determining Ventilation Rates . . . . . . . . . . . . . . . . . . . . . 95
9.3 Air Handling Unit 2 – Supply and Return Air System . . . . . . . . . . . . . . . 97
9.3.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
9.3.2 Sequence of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
9.3.3 Operation Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
9.4 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
9.4.1 Determination of Ventilation Rate Requirements . . . . . . . . . . . . . . . . 98
9.4.2 Comparison with the Actual Ventilation Provided . . . . . . . . . . . . . . . . 99
9.4.3 Measurement of CO2 Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
9.5 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9.6 Exhaust Fans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9.6.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
9.6.2 Operation of the Exhaust Fan . . . . . . . . . . . . . . . . . . . . . . . . . 101
9.7 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
9.7.1 Exhaust Provided vs. Exhaust Requirements . . . . . . . . . . . . . . . . . 101
9.7.2 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
9.8 Air Handling Units 4 and 5 – Supply Air Only Systems . . . . . . . . . . . . . 105
9.8.1 Sequence of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
9.8.2 Originial Operation Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
9.9 Method of Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

vi
 9.9.1 Verification of Operation Sequence . . . . . . . . . . . . . . . . . . . . . . 106
9.9.2 Cost of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
9.9.3 Analysis of Space Serviced . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
9.10 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
9.10.1 Actual Occupancy Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
9.10.2 Total Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
9.10.3 Impact of Reductions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
9.10.4 Winter Month Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
9.10.5 Summer Month Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
9.11 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
9.11.1 Reduced Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
9.11.2 Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
9.11.3 Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
9.11.4 Payback Period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
9.12 Air Handling Units 1 and 3 – Supply and Return with Mixed Air Systems . . 119
9.12.1 Sequence of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
9.12.2 Original Operation Schedule . . . . . . . . . . . . . . . . . . . . . . . . . . 120
9.13 Method of Investigatoin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
9.13.1 Verification of Operation Sequence . . . . . . . . . . . . . . . . . . . . . . 121
9.13.2 Cost of Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
9.13.3 Analysis of Space Served . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
9.14 Findings for Air Handling Units . . . . . . . . . . . . . . . . . . . . . . . . . . 125
9.14.1 Understanding Air Handling Unit 1 . . . . . . . . . . . . . . . . . . . . . . 125
9.14.2 Actual Occupancy Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
9.15 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
9.15.1 Scheduling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
9.15.2 Reducing Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
9.15.3 Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
9.15.4 Payback Period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
9.16 Findings for Air Handling Units 3 . . . . . . . . . . . . . . . . . . . . . . . . . 129
9.16.1 Understanding Air Handling Unit 3 . . . . . . . . . . . . . . . . . . . . . . 129
9.16.2 Variable Air Volume Boxes . . . . . . . . . . . . . . . . . . . . . . . . . . 130
9.16.3 Actual Occupancy Load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.17 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.17.1 Temperature Model Solution . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.17.2 Reducing Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
9.17.3 Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
9.17.4 Payback Period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133

10.0 Lighting in Maxwell Dworkin . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

10.1 Analysis of the Current Lighting Situation and Proposal to Improve Efficiency . 137
10.1.1 Light Measurements and Units . . . . . . . . . . . . . . . . . . . . . . . . . 137
10.1.2 Light Bulb Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
10.1.3 Lighting Contrast Levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
10.1.4 Summary of Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.2 Lighting Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.2.1 Ground Floor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
10.2.2 Faculty and Staff Offices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
10.2.3 Hallways and Common Areas . . . . . . . . . . . . . . . . . . . . . . . . . 143
10.3 Summary of Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

vii
11.0 Generalization to Other Harvard Buildings . . . . . . . . . . . . . . . . . . . 145
11.1 Applicability to William James Hall and the Science Center . . . . . . . . . . . 147
11.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
11.1.2 Ventilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
11.1.3 Lighting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
11.1.4 Additional Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
11.2 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

12.0 Summary of Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

12.1 Overall Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153

Appendices
Appendix 1 Temperature Data for Boston, MA
Appendix 2 Energy Balance Calculations for Maxwell Dworkin
Appendix 3 Conversion Calculations from Energy Source to Acres of Forest
Appendix 4 Utility Information for Maxwell Dworkin
Appendix 5 Calculations and Measurements for AHUs 4 and 5
Appendix 6 Calculations and Measurements for AHUs 1 and 3
Appendix 7 Schematic Drawings of the Operation of AHU’s 1, 2, 3, 4, & 5
Appendix 8 BOCA International Mechanical Code Ventilation Requirements
Appendix 9 Decora Wall Switch Occupancy Sensor
Appendix 10 Lighting Tables from the Illuminating Engineering Society of North America
Appendix 11 Extech Light-Meter Profile
Appendix 12 Lighting Measurements for a South-Facing Office Space on a Sunny Day
Appendix 13 Lighting Measurements and Calculations
Appendix 14 Telaire® 7001 Carbon Dioxide and Temperature Monitor

viii
ix
x
ES 96 Spring 2002

1.0 INTRODUCTION

1
ES 96 Spring 2002

2
 ES 96 Spring 2002

1.0 Introduction

This spring, the Engineering Sciences 96 design seminar of 11 students was presented with the task of
investigating in detail the mechanical systems of one of Harvard’s newer buildings. After gaining a
full understanding of the advanced systems involved, we used our research of these systems to offer
recommendation on how they could be more efficiently operated with greater savings to the
university and less impact on the environment.

Efficient use of Harvard’s buildings is an important concern today and one that will become
increasingly relevant in years to come as Harvard’s buildings continue to age. The buildings on
campus range anywhere from newly built to more than 250 years old, which means that the
mechanical systems that run them are as diverse as Harvard’s student body itself. By measuring a
building’s efficiency based on the cost to run the building per sq. foot, we created an efficiency
comparison of several Harvard buildings. Though it may seem counterintuitive, many of Harvard’s
older buildings may actually be more efficient than those built today. Newer buildings offer many
modern conveniences that the buildings of the past did not and those comforts equal more money.

When undertaking our project, we decided it would be best to focus on one building. After choosing
a building, we could determine if our insights of how to operate that building more energy efficiently
could be generalized to buildings such as the Science Center and William James Hall. We therefore
chose to first study Maxwell Dworkin, as it provided the ideal setting for gathering our initial
research. Maxwell Dworkin is one of Harvard’s newest buildings, completed in 1999, and a friendly
staff and faculty were willing to assist in our experimentation, and this made it an attractive choice. It
hasa sophisticated and accessible mechanical system that is equipped with several hundred control
points which can be accessed and managed by a central control program. The class then toured the
mechanical rooms of Maxwell Dworkin to get a first-hand look at its mechanical, electrical, and
heating ventilation air-conditioning (HVAC) systems. While touring these systems, we questioned
how FAS buildings are provided with heating and cooling.

3
 ES 96 Spring 2002

The answer? Buildings are heated via an underground utility tunnel to a steam and electrical
generating plant at Western Avenue and Memorial Drive on the Charles River. The underground
tunnel carries steam under pressure, electrical lines, and communication lines directly to buildings.
Chilled water for cooling, however, is generated in the basement of the Science Center and is
distributed directly to buildings through buried insulated pipes.

The electricity, steam and chilled water then meet the utility room. Steam energy is converted to hot
water and pumped throughout Maxwell Dworkin to provide heating. Steam flow is metered by the
flow rate of the steam condensate which is returned to the power plant. Chilled water is also pumped
throughout the building and then returned to the plant slightly warmer than it arrived at Maxwell
Dworkin. Billing for chilled water is metered by the flow rate of the water times the temperature rise
in the water returned to the plant. Electrical power enters the building through a separate vault and is
there metered for billing purposes.

Armed with this knowledge, we were able to identify three major sources of energy consumption in
Maxwell Dworkin–lighting, air handling and fan coil units, and chilled and hot water usage. We then
split into smaller groups that would address each of these areas. Each group looked at energy
consumption records, and performed individual data records such as measuring and recording CO2
levels over time, measuring light levels, and looking at the time history of energy use in rooms and
lecture halls. From that data each group developed mathematical models to predict room conditions
under different operating settings. This information was then used to determine how each of the
systems could more efficiently and environmentally be managed. These insights of how to operate
the building more energy efficiently were used to determine if improvements might be made to the
Science Center and William James Hall, though this is by no means an exhaustive list of the buildings
that could be aided by such improvements.

It is imperative that Harvard do its part in helping the environment. We hope to help people
understand how greenhouse emissions may be decreased for the operation of every Harvard building
while maintaining the building as a pleasant place to teach, go to classes, and do research. This paper
presents our findings for the potential environmental and monetary savings in lighting, air handling
and fan coil units, and chilled water usage along with the background knowledge necessary to
understanding the environmental impact.

4
 ES 96 Spring 2002

2.0 ENERGY & ENVIRONMENTAL INDICATORS

5
ES 96 Spring 2002

6
 ES 96 Spring 2002

2.1 Energy, Economy, and Environment

Measures of economic and environmental efficiency often take the same form: the consumption of
fewer resources to achieve equal provision of products and services makes good business and
ecological sense. This is particularly true with respect to energy saving endeavors: conservation often
pays a commendable double dividend. Unfortunately, where the economic dividend is easily
understood and measured in dollars, the environmental benefits are sometimes harder to calibrate.
This section discusses the motivation behind calculating an environmental dividend and provides
detailed information on the calculations that undergird the environmental indicators of the analysis.

2.1.1 Environmental Issues and Opportunities

The April 2001 edition of the Environmental Building Network’s Newsletter stated that commercial
buildings in the United States account for 36% of all energy consumption in the United States – an
astronomical 3.6 billion MWh per annum. Of this amount, these buildings – which include private
and public offices, as well as retail space – account for 62% of all electricity use and 30% of all
greenhouse gas (GHG) emissions. As such, commercial buildings represent a tremendous
opportunity to improve the environmental efficiency of the United States. Concentrated efforts in the
design and retrofit of building with advanced technology are important steps forward in capitalizing
upon this opportunity. Yet in order to move forward, an easily understandable means of comparing
the benefits to costs is necessary.

7
 ES 96 Spring 2002

2.1.2 Indicators of Energy and Environmental Efficiency

The lack of easily comparable and standardized indicators often renders discussions of environmental
efficiency contentious and frustrating. The scale of the factors involved – billions of dollars, 3.6
billion MWhs – are often confounding. In order to avoid unnecessary confusion and facilitate
comprehension of the suggestions made in this report, we have decided to use four indicators
throughout this analysis:
• Dollars
• Operational Cost per Square Foot
• Units of Energy or Utility
• Acres of US Commercial Forest

The first three indicators, in our estimation, are relatively self-explanatory. For each type of
operation or utility analyzed, estimated savings will be presented in economic terms. For example,
the changes proposed to the lighting scheme have the potential to save $13,000. Moreover, the
decreases to the operational cost per square foot will be presented; using the lighting example,
changing the lighting schema in Maxwell Dworkin could reduce the operating cost per square foot
from $2.73 to $2.54. Furthermore, savings will be presented in terms of the units of the energy or
utility consumed.

Our fourth indicator, in contrast, requires further explanation. Recent concern over global climate
change has prompted heated international discussion over the role of greenhouse gases (GHG), which
change the heat-absorbing properties of our atmosphere. Governments, institutions and companies
are under increasing pressure to calibrate and reduce the amount of carbon dioxide, methane, and
sulfur hexafluoride. One of the recent initiatives in this debate has been to consider forests as “carbon
dioxide sinks” that sequester GHGs, and thereby compensate for the GHG emitted as a result of use
of fossil fuel energy. As such, we decided to use the reduction in the number of acres of US
commercial forest required to absorb Maxwell-Dworkin's carbon dioxide emissions as our fourth
indicator.

8
 ES 96 Spring 2002

2.2 Acres of Forest as an Environmental Indicator

Embedded in these discussions, however, are assumptions as to the efficiency of energy production.
The variety of methods used in the production of energy requires that any serious discussion of GHG
reductions measures make these assumptions explicit to avoid unnecessary confusion. In recognition
of this necessity, we analyzed the carbon dioxide production involved in the three major utilities our
study focused on. After identifying Electricity, Chilled Water and Steam as the principal targets
for reduction, data accumulation on the efficiency of energy production from each of these three
utilities was gathered in order to calculate the greenhouse gas emissions. Furthermore, research from
the Naval Research Laboratory1 revealed that 1.19 tons of carbon dioxide are annually absorbed by 1
acre of US Commercial Forest per annum.

2.2.1 Electricity to Acres of Forest

Electricity at Harvard is provided by the NSTAR Corporation, Massachusetts’ largest investor owned
electric and gas utility. Recent corporate documentation demonstrated that the electricity that
Harvard consumes is generated in a variety of different ways (Figure 2.1).

Figure 2.1 Fuel mix of electricity purchased by Harvard University

1
http://itest.slu.edu/articles/90s/hannan.html

9
 ES 96 Spring 2002

The large proportion of nuclear energy in Harvard’s energy supply complicates the calibration of
GHG emissions from electricity use, however. As a result, we decided to use the EMISS program
provided by the National Institute of Standards and Technology, which gives the conversion factor to
carbon dioxide by the different region in the United States, taking into account the different fuel mix.
EMISS stated that 234.6 kg of carbon dioxide are emitted per MMBtu of electricity produced. The
calculations in Appendix 3 reveal that this means 1350 kWh of electricity would need to be reduced
in order to reduce the area of forest required to absorb the emissions by 1 acre of US Commercial
Forest.

2.2.2 Chilled Water to Acres of Forest

The chilled water used to cool Harvard’s building is generated at the Science Center from 5 distinct
chillers of differing capacities. Maxwell Dworkin is billed for it’s chilled water use in ton-days, a
unit of energy equivalent to what would need to be removed from 1 ton of water at 32 degrees
Fahrenheit to generate 1 ton of ice at the same temperature. In more quantitative terms, a ton-day
represents the same amount of energy that a 12,000 BTU/hr engine would output if it ran for twenty-
four hours. Our research revealed an Energy Efficiency Ratio (EER) of 12 for the plant: that is, it
takes 1000 watts to produce a ton (12,000 BTU) of cooling. As such, our calculations in Appendix 3
reveal that a reduction of 56 ton-days of cooling would reduce the area of forest required to absorb
the GHG emissions by 1 acre of US Commercial forest.

2.2.3 Steam to Acres of Forest

Finally, steam at Harvard is co-generated at the Commonwealth Electric Company on Memorial

Drive from a highly viscous fuel known as “Bunker Oil.” This steam circulates throughout the
campus from September to April, providing the energy required for heating water, which in turn heats
the buildings themselves. Using information from the Department of Energy with respect to the
energy content of “Bunker Oil,” a standard chemical equation for the combustion of hydrocarbons 2,

CxHy + vO 2(O2 + 3.76N 2) Æ vCO 2CO2 + vH 2OH 2O + vN 2 N 2 (Eq. 2.1)

†
2
Equation of Combustion from Sonntag et al., Funadmentals of Thermodynamics, p 524, Wiley:New
York:1998

10
 ES 96 Spring 2002

and an assumed value of the plant’s efficiency of 80%3, we concluded that a reduction of 12.5
MMBTU of steam consumption is required to reduce the GHG absorption requirement by 1 acre.

3
Provided by Professor Frederick H. Abernathy, Abbott and James Lawrence Professor of Engineering and
Gordon McKay Professor of Mechanical Engineering, Division of Engineering and Applied Sciences,Harvard
University

11
ES 96 Spring 2002

12
 ES 96 Spring 2002

3.0 ENIVRONMENTAL IMPACTS

13
ES 96 Spring 2002

14
 ES 96 Spring 2002

3.1 Environmental Impacts of Maxwell Dworkin

Realizing the potential impact of buildings on the environment, our class explored these implications
to Harvard University. On the Cambridge campus alone, Harvard possesses over two hundred
buildings. Additionally, buildings are constantly being renovated and new structures continue to be
built. By understanding the impact on the environment of current buildings, future construction
projects can incorporate this knowledge into their plans.

Specifically, our class chose to examine Maxwell Dworkin. Our choice was not made without
reason; Maxwell Dworkin is one of the most modern building on campus. Its complex operating
systems were designed to be energy efficient. Our hope was that having investigated the opportunity
for energy savings in a modern building on campus, we could generalize our findings to other
buildings. By examining the records of Maxwell Dworkin’s utility use, we calculated a total utility
budget and its equivalence in acres of U.S. Commercial Forest that would have been required to
absorb the carbon dioxide emissions from typical utility plants generating the utilities. We chose the
utility use for the fiscal year 2001 because it was the most complete data set obtained from Harvard
University Operations Services. A projected budget was computed using the utility costs for the
Fiscal Year 2003 alongside the utility use for the year 2001. These costs include distribution costs for
the specific utility (Table 3.1). Refer to Appendix 4 for detailed data. The calculated budget includes
the following utilities: chilled water, steam, electricity and water. With these four utilities, the
estimated utility cost for Maxwell Dworkin is $305,816.

15
 ES 96 Spring 2002

TABLE 3.1 Projected Budget of Maxwell Dworkin

Fiscal Year 2001 Utility Usage Cost of Utility in Dollars for
Fiscal Year 2003**
Electrical (kWh) 1153728 x 0.1095 dollars/kWh = 126333.2
Steam (MMBtus) 3753 x 17.97 dollars/MMbtu = 67441.41
Chilled Water (Ton-Days) 12540 x 8.7 dollars/Ton-Day = 109098
Water (Ccf) 290 x 10.15 dollars/Ccf = 2943.5
$ 305816.1 Total Cost
*Utility Usage obtained from Harvard University Operations Services.
**Utility Cost for the Fiscal Year 20034

In order to obtain an equivalent measure in acres, the emission rate for each utility was tabulated
separately using the calculated conversions explained above in Section 2.2. See Table 3.2 for
calculations. The total acreage required to compensate for the Maxwell Dworkin’s utilities for the
fiscal year 2001 is 1379 acres.

TABLE 3.2 Environmental Impact of Maxwell Dworkin

Fiscal Year 2001 Utility Usage* Conversion to acres Acres
Electrical (kWh) 1153728 / 1350 kWh/acres = 854.6
Steam (MMBtus) 3753 / 12.5 MMBtus/acre = 300.2
Chilled Water (Ton-Days) 12540 / 56 Ton-Days/acre = 223.9
1378.7 Total acres
*Utility Usage obtained from Harvard University Operations Services.

To place this number into perspective, our class found it useful to compare this acreage to a more
familiar piece of land, specifically Harvard Yard. The entire area of Harvard Yard enclosed by the
black iron gates, as shown in Image 3.1, equals 22 acres. In comparison to the total acres needed to
compensate for Maxwell Dworkin’s carbon dioxide emissions, it would require approximately 63
Harvard Yards.

These calculations motivated our class to examine more energy efficient ways in which to operate
buildings at Harvard University. If just one building produced this enormous amount of carbon

4
FAS Planning numbers for Fiscal Year 2003. Courtesy of the Office of Physical Resources.

16
 ES 96 Spring 2002

dioxide emissions, imagining the impact of all Harvard University buildings is astounding. Clearly, it
is imperative for Harvard University to consider its impact on the environment.

IMAGE 3.1 Map of Harvard Yard5

3.2 Energy Use in Maxwell Dworkin and Harvard Buildings

Having decided to focus our study on Maxwell Dworkin, it was important to compare the relative
energy efficiencies of various buildings. By using current utility planning costs for the fiscal year
2003, we compared the utility cost per square foot of various buildings. The utility usage numbers
are from the fiscal year 2001. Keeping in mind that Maxwell Dworkin was built to be energy
efficient, its success is limited (Figure 3.1). See Appendix 4 for detailed utility data.

5
Image obtained from Harvard University Map http://map.harvard.edu/level2/2Yard.shtml

17
 ES 96 Spring 2002

Utility Cost per Square Foot

$5.41
$4.99

$3.74

$2.73
$2.39

Maxwell Pierce Hall William Science Gordon

Dworkin James Center McKay

FIGURE 3.1 Utility cost per square foot of 5 buildings on the Harvard Campus

In comparison to William James Hall, the Science Center and Gordon McKay, Maxwell Dworkin
appears to be more energy efficient. It is important to note that these buildings have different
functionalities. William James Hall, the Science Center and Gordon McKay all possess laboratories
that heighten utility expenses. However, Pierce Hall and Maxwell Dworkin share similar
functionality. Both buildings contain mostly offices and lecture halls. When compared to Pierce
Hall, Maxwell Dworkin is more expensive to run. Pierce Hall is an older, basically un-insulated
building while Maxwell Dworkin is a modern, well-insulated building. This puzzling result forced
our class to further examine the exact sources of energy consumption within Maxwell Dworkin.

3.3 Energy Consumption in Maxwell Dworkin

By examining the monthly trends of utility costs, our class gained an understanding of how the
building consumes energy throughout the year. The three utilities trended were electricity, steam and
chilled water. The data used was from January 2000 to April 2002. See Figures 3.2, 3.3, and 3.46.

6
Utility Usage obtained from Harvard University Operations Services

18
 ES 96 Spring 2002

Monthly Trends of Electricity

120,000

100,000

80,000
KWh

60,000

40,000

20,000

0
ch

ch

ch
ay

ay
y

y
ov er

Ja r

ov er

Ja r
Se July

Se July
be

be
ar

ar

ar
b

b
ar

ar

ar
M

M
nu

nu

nu
em

em

em

em
M

M
Ja

pt

pt
N

N
Years 2000-2002

FIGURE 3.2 Monthly Trends of Electricity

The electric use throughout the years remains relatively constant, which corresponds to a constant use
of items such as lights, computers, elevators, etc. This result seems reasonable since professors and
graduate students work all year round.

The steam trends also appear logical. During the winter months when the most heating is required,
the most steam is being used. Additionally, as the steam is shut during the summer months, and there
is no steam usage during the appropriate period.

19
 ES 96 Spring 2002

Monthy Trends of Steam

1000
900
800
700
MMBTUs

600
500
400
300
200
100
0
ch

ch

ch
ay

ay
y

y
ov er

Ja r

ov er

Ja r
Se July

Se July
be

be
ar

ar

ar
b

b
ar

ar

ar
M

M
nu

nu

nu
em

em

em

em
M

M
Ja

pt

pt
N

N
Years 2000-2002

FIGURE 3.3 Monthly Trends of Steam

The third trended utility was chilled water. As expected, chilled water usage peaks in the summer,
when most cooling is needed. However, chilled water costs do not vanish during the winter months.
In fact, during the winter months they are nearly one-third as much as during the summer months.
Maxwell Dworkin needs cooling all year round. Furthermore, the building is being simultaneously
heated and cooled during the winter. These unusual patterns of utility usage needed further
investigation.

20
 ES 96 Spring 2002

Monthy Trend of Chilled Water

2500

2000
Ton-Days

1500

1000

500

0
ch

ch

ch
ay

ay
y

y
ov er

Ja r

ov er

Ja r
Se July

Se July
be

be
ar

ar

ar
b

b
ar

ar

ar
M

M
nu

nu

nu
em

em

em

em
M

M
Ja

pt

pt
N

N
Years 2000-2002

TABLE 3.4 Monthly Trend of Chilled Water

We needed to determine the demands of the building that created this pattern of usage. Having
realized the impact of Harvard buildings on the environment, our class had analyzed the sources
behind the impact. However, our investigation had only proved that Maxwell Dworkin’s building
operations required a greater understanding in order to form a complete picture.

21
ES 96 Spring 2002

22
 ES 96 Spring 2002

4.0 BUILDING ENERGY BALANCE

23
ES 96 Spring 2002

24
 ES 96 Spring 2002

4.1 Building Scale Energy Balance

Knowing the utility usage for the building, we placed these factors alongside other energy
components to create an energy balance of the entire building. Our class wanted to determine the
forms of energy flowing in and out of the building during various times of the year. We initially
came up with a very basic energy balance for the building. See Figures 4.1a and 4.1b for December
Energy Balance. See Figures 4.2a and 4.2b for July Energy Balance. Please refer to Appendix 2 for
calculations.

Sunload Heat
6% Transfer
through
Envelope
21%
Electrical
39%
Ventilation
47%

Steam
55% Chilled
Water
32%
FIGURE 4.1a December Energy In FIGURE 4.1b December Energy Out

During the winter months, we calculated energy coming into the building from three major sources.
The first was sunload through the building’s exterior walls and windows. The electrical load from
such devices as computers, lights and elevators, accounted for thirty-nine percent of the energy load
into the building. The most significant factor bringing heat and energy into the building is steam,

25
 ES 96 Spring 2002

which accounted for fifty-five percent. Energy is carried out of the building by heat transfer through
the exterior building frame, chilled water and ventilation. During the winter months, ventilation
systems of the building are responsible for nearly fifty-percent of the energy exiting the building.
These winter calculations, however, proved to be problematic for the class. There is a nearly twenty-
five percent of energy flowing in unmatched by the energy leaving the building. This discrepancy is
most likely due to unaccounted energy loss due to exterior frame radiation, and inaccurate weather
data.

Heat
Transfer
through
Ventilation Envelope
12% 1%

Electrical
Sunload 53%
35% Chilled
Water
99%
FIGURE 4.2a July Energy In FIGURE 4.2b July Energy Out

As for the summer months, the energy coming into the building is from the sunload, the electrical
load, and ventilation. Since the steam transmission into the tunnels is stopped, there are no steam
costs. Chilled water and heat transfer are the means that energy escapes the building. Our July energy
balance came much closer to accounting for energy flow within Maxwell Dworkin. Only ten percent
of energy in was left unmatched by energy leaving the building. Although neither energy balance
was perfected, they were able to point out major areas of energy sources and sinks.

These inadequate balance of energy made it clear to our class that further investigation was required
to understand, on a more fundamental level, the heat exchange and energy flow within the building.
In order to be able to tackle all aspects of the building in depth, our class divided into separate groups.
Each group was given a specific area to research and analyze. In this manner, our class improved

26
 ES 96 Spring 2002

upon this energy balance and found other causes for heat loss and gains within the building. After
gaining a perspective of the potential environmental impact of Harvard buildings, our class began
uncovering the basic operating features of Maxwell Dworkin. Armed with puzzling data and
unsolvable problems, we continued our investigation by delving deeper into individual components of
the building, including, an office temperature model, ventilation systems and the lighting scheme.

27
ES 96 Spring 2002

28
 ES 96 Spring 2002

5.0 INTELLIGENT CONTROL OF

BUILDING SYSTEMS

29
ES 96 Spring 2002

30
 ES 96 Spring 2002

5.1 APOGEE: The Intelligent Control System

5.1.1 System Description

Siemens Business Technologies and APOGEE Protocol Communication have collaborated to develop
the “premier integrated building management solution available”7 - a control network that digitally
operates the components within its building. Commonly referred to as “APOGEE”, this unit provides
a modernized Maxwell Dworkin with more control functionality at more control points in a more
sophisticated and unified manner. Technologically, APOGEE runs as follows: “data is transmitted
and received over a dedicated set of wires running throughout the building. Each device on this
network is assigned an address and can be accessed by the host automation controller.”8 Users can
access any type of information passed over this bus to monitor, schedule, update, review, and easily
operate its building's complex systems.

Ease of data accessibility is one of APOGEE’s strongest features. According to Siemens, “through
[APOGEE’s] powerful, real-time management workstation, information and control is always
accessible to those who need it – when and where they need it.”9 Users can connect to APOGEE
through a corporate network, through any system panel, and even through a basic modem. Users can
also send system and trend reports to other users on the corporate network by taking advantage of
APOGEE’s InfoCenter package. Siemens sums up such broad capabilities in its mantra on APOGEE
access: “Sit anywhere, do anything”.10

7
Quote taken from HVAC Controls section on Siemens website (www.sbt.siemens.com).
8
Quote taken from Reliance Electric Publication D-2976 July 2000 (www.reliance.com).
9
Quote taken from HVAC Controls section on Siemens website (www.sbt.siemens.com).
10
Ibid.

31
 ES 96 Spring 2002

IMAGE 5.1 APOGEE Control System

5.1.2 Method of Investigation

For a preliminary understanding of the system, we researched the APOGEE control network online.
To clarify any misunderstandings of the description that we had after our online research and to
experience APOGEE first-hand, we met with Maxwell Dworkin’s Siemens contact, Scott Gaines.
Scott demonstrated the ease and flexibility of the system and also allowed us to access Maxwell
Dworkin’s control data through the APOGEE network. Finally, we spoke to the managers of the
building who work with Maxwell Dworkin’s APOGEE system on a daily basis. Ed Jackson, Greg
Kousidis, and Steve Robichaud all shared with us the details of this intricate relationship between the
particular building and APOGEE.

By the end of our quest to comprehend APOGEE, we felt that we had a solid understanding of the
way in which such a network controls and operates a complex Maxwell Dworkin.

32
 ES 96 Spring 2002

5.1.3 Findings

The APOGEE System provides scheduling functionalities so that the building components need not
be manually turned on and off, one-by-one. Such functionalities also ensure that these components
need not run excessively.

If the building manager, for instance, is aware that Maxwell Dworkin's G115 lecture hall will not be
used at all on Tuesday morning, he can pre-program the APOGEE System to leave that specific room
in "Unoccupied" mode so that the room's automated components can rest during that vacant period. In
this way, APOGEE - when used efficiently - can be an important energy saving tool.

APOGEE can also be a tool that provides comfort to its occupants, as it monitors comfort levels such
as temperature. Operating Maxwell Dworkin's heating, cooling, ventilating, and recirculating devices,
APOGEE keeps the building's temperature with the pre-set temperature range.

5.1.4 Recommendations

Through researching APOGEE, it is apparent that this control network has the potential to optimize
Maxwell Dworkin’s energy efficiency, quality of control, and information management. It is now
necessary to reexamine the APOGEE control commands and parameters that are currently operating
Maxwell Dworkin in order to ensure efficient use of the building’s Heating, Ventilation & Air-
Conditioning (HVAC) system.

It is therefore important to consider APOGEE in every aspect of this project’s goals. Maxwell
Dworkin’s energy problem is discussed in many arenas throughout this paper, from an adaptable
temperature model to lighting efficiencies. Though not necessarily mentioned directly, any
recommended change to Maxwell Dworkin’s digital control system translates to a change to the
APOGEE network. In this way, APOGEE’s accessibility, flexibility and information management all
play significant roles in the realization of this project’s numerous findings and recommendations.

33
 ES 96 Spring 2002

5.2 Air Ventilation and Circulation Systems Controlled by APOGEE

5.2.1 Air Handling Units

At the heart of the building’s HVAC operations is the air-handling unit (more commonly referred to
as the AHU). The fundamental role of an AHU is to take in air from outside and condition it through
heating or cooling until the inside air achieves a desired range. Despite this apparently simplistic
fundamental role, AHUs can vary in many levels of complexity.

Diagram 5.1 displays a basic AHU that simply takes in outside air and passes it along the air duct
through a filter, into the supply fun, and then eventually to the appropriate destination space.

DIAGRAM 5.1 Basic Air Handling Unit

Diagram 5.2 is a more complex AHU design that conditions air brought in with heating or cooling
until the air has achieved the temperature specified by APOGEE control. Once the air reaches the
supply air duct, it will pass through either an activated heating or cooling device within the system. If
the air’s temperature is above the set-point temperature for the area, an APOGEE command activates
the cooling coil to cool the air. Conversely, if the air’s temperature is below the set-point
temperature, an APOGEE command activates the heating coil to heat the air. The newly conditioned
air is then sent to its destination space.

34
 ES 96 Spring 2002

DIAGRAM 5.2 Air Handling Unit with heating and cooling devices

Diagram 5.3 builds on the previous AHU design by incorporating mixed air, an important energy-
saving tool. As opposed to exhausting 100% of air returned from rooms in its building, this system
recycles a portion of the return air to mix with the incoming outside air. This mixed air is often at a
more moderate temperature than the outside air, thus it requires less energy to condition.

DIAGRAM 5.3 Air Handling Unit with heating and cooling devices, incorporating mixed air

35
 ES 96 Spring 2002

There are five AHUs ventilating Maxwell Dworkin. AHU #1 is the HVAC system for lecture hall
G115; AHU #2 controls the basement; AHU #3 controls the corridors and remaining rooms (aside
from G115) on the ground floor; AHU#4 & AHU #5 controls the systems on Floors 1, 2, and 3. The
power required to operate and control these five units, as well as the chilled and hot water needed to
condition the building’s air all contribute to the building’s total energy use.

5.2.2 Fan Coil Units

While Air Handling Units bring conditioned air to designated spaces, fan coil units recirculate air
within their room. A fan coil unit is a radiator that has hot and/or cold water to re-heat or re-cool the
air already conditioned by the AHUs.

Maxwell Dworkin uses top-of-the-line four-pipe fan coil units with heating and cooling coils that can
run year-round. Such a design differs from Pierce Hall’s “seasonal” system, for example, where there
is only one device that can serve as either heating or cooling at any one time.

With Maxwell Dworkin’s four-pipe fan coil units, as demonstrated below by Diagram 5.4, air is taken
into the system through a fan operated by APOGEE, passing by either an activated heating or cooling
coil. This “active” coil status is determined by APOGEE, which is connected to all fan coil unit
thermostats. If the room thermostat is above the set-point thermostat range, an APOGEE command
activates the cooling coil to cool the room and closes the heat coil. If the room thermostat is below
the thermostat set-point range, an APOGEE command activates the heating coil to heat the room and
closes the chilled water supply to the unit. This air is sent back out into the same room at a newly
conditioned temperature and with a different dew point.

36
 ES 96 Spring 2002

DIAGRAM 5.4 Fan Coil Unit

Maxwell Dworkin is equipped with fan coil units in every office space and research area
on Floors 1, 2, & 3. The power required to operate and control these units, as well as the chilled and
hot water needed to condition the building’s air all contribute to the building’s total energy use.

37
ES 96 Spring 2002

38
 ES 96 Spring 2002

6.0 OFFICE ENERGY BALANCE

39
ES 96 Spring 2002

40
 ES 96 Spring 2002

6.1 Adaptable Temperature Model

Although construction of an energy balance for an entire building is useful, any recommendations
drawn from such a macroscopic view run the risk of being too general. In other words, what may be
a good recommendation for the building as a whole may be a poor recommendation for individual
offices or laboratories. In order to make temperature recommendations for each of these spaces, it is
essential to understand them thermodynamically. We demonstrate a thermodynamic understanding of
spaces in Maxwell Dworkin through the development of an adaptable temperature model. This model
will predict the temperature in a room over time and is adaptable to any closed space similar to the
offices and laboratories in Maxwell Dworkin.

In this section, we will discuss first the existing temperature regulation for a typical room in Maxwell
Dworkin and then other uncontrollable factors that affect temperature. Although there is a range of
temperatures within a room, for the purposes of our discussion we will define the temperature of the
room to be the air temperature next to the thermostat. This assumption is particularly valid when the
air in the room is circulated, but even at night, when stratification of the air occurs, we find the
temperature strays fewer than two degrees from this value.

6.1.1 The Status Quo

Temperature regulation in Maxwell Dworkin is under the control of the Siemens APOGEE system.
Specifically, in office and research spaces, temperature regulation is achieved by the fan coil units
(FCUs). The FCUs internally circulate either heated or chilled water to condition the air within the
space. The FCUs are active between 7am and 7pm, with the possibility of a three-hour override of
the schedule if there is occupancy outside of this timeframe. When active, the FCUs heat or cool
based on thermostatic control.

Such thermostatic control in Maxwell Dworkin currently takes the form of what we have called a
narrow-band constraint. The thermostat has a static setpoint—generally 72°F—and a “dead band”

41
 ES 96 Spring 2002

two degrees above and two degrees below that setpoint. Thus, with a 72°F setpoint, the room is
considered appropriately conditioned as long as the temperature remains between 70°F and 74°F. If
the temperature drifts out of this range, active heating or cooling occurs. Of course, individuals
occupying office and research space can adjust the thermostat setpoint, but the APOGEE system does
impose absolute limits.

6.2 Method of Investigation

To assess whether or not we could improve upon the narrow-band constraint strategy currently in
place, we chose to develop a flexible mathematical model that would allow us to simulate the
behavior of an office or research space. With such a tool, we could assess the amount of savings
possible with a proposed control strategy and ensure that it did not compromise the comfort level of
the space.

6.2.1 Background

Before building such a model, some understanding of thermodynamics is necessary. Within a room,
there are several sources and sinks of heat. The largest of these inputs are due to the electrical
load—lights, computers, monitors, and other appliances all produce heat when active—and the
sunload, especially in the case of south-facing office windows. Active FCU heating and human heat
emissions also serve to warm a room, but to a much lesser extent than the other sources, as depicted
in Figure 6.1. The majority of heat removal is in the form of active cooling from the FCUs.
Additional heat loss also occurs through closed windows (via transfer through the glass and leakage
through the seals), although this is more than balanced by the electrical load and sunload. It should
be noted that passive transfer can also be a heat gain, dependent on the relative temperatures between
the interior and exterior of the building. For example, if it is warmer outside a window than inside,
heat will be transferred into a room, causing a rise in temperature to occur.

42
 ES 96 Spring 2002

FIGURE 6.1 Energy balance for a south-facing office with one-person

occupancy on a sunny April morning. Chart assumes thermal equilibrium (i.e.
constant temperature) for the room.

6.2.2 Heat Capacity of a Room

Another major factor involved in the temperature characteristics of a room is the heat capacity of the
room itself. A homogeneous material requires some finite, characteristic amount of heating to cause
its temperature to rise by one degree Fahrenheit. This is the material’s specific heat—for our
purposes measured in units of BTU/(lb-°F)—and is mass invariant. Thus, the specific heat for
aluminum will be the same no matter what the mass of the sample under investigation. However,
when the specific heat is multiplied by the mass of a particular sample of the material, the value
obtained is mass dependent and is called the heat capacity of that sample, in units of BTU/°F. In the
case of modeling the temperature characteristics of an entire room, a conglomerate heat capacity
value, Cr, would be most useful. Fortunately, this can be calculated by summing the individual heat

43
 ES 96 Spring 2002

capacities of all the materials comprising the room. Once Cr is known, the rate of temperature change
of the room and the rate of heat gain for that room are related by Equation 6.1:

dTr (t )
Cr = Qr (t ) (Eq. 6.1)
dt

where Tr(t) is the room temperature at time t, and Qr(t) is the rate of heat gain to the room.

To attempt to estimate the heat capacity of a room, we performed the following inventory of the
contents of a faculty office Table 6.1. One can quickly see that the major contributors to Cr are from
the walls, floor, and ceiling, all of which are well insulated and thus designed to retain heat within the
room.

TABLE 6.1 Specific Heats for Common Exposed Surfaces in a Room

Item Mass in lbs Volume in ft3 Specific Heat in BTU/(lb-°F)† [material]
Books 400 0.4 BTU/(lb-°F) [wood]
Shelves 50 0.4 BTU/(lb-°F) [wood]
Desk 200 0.2 BTU/(lb-°F) [stone, plastic]
Chair 25 0.2 BTU/(lb-°F) [plastic]
Air in room 200 0.25 BTU/(lb-°F)
Walls (exluding 100* 13 BTU/(ft3-°F) [gypsum board]
windows, door)
Floor 82* 23 BTU/(ft3-°F) [concrete]
Ceiling 82 2.6 BTU/(ft2-°F)
Window glass‡ 2.5 0.20 BTU/(lb-°F)
Door 150 0.4 [wood]
* Assuming four-inch heat penetration depth
† Sources: Bobenhausen, William, Simplified Design of HVAC Systems;
www.ac.wwu.edu/~vawter/PhysicsNet/Topics/Thermal/HeatCapTable.html
‡ Assuming 120 ft2 area and 0.25 inch thickness. We ignore the glass due to its negligible total volume.

A final factor to consider is that heat capacity only considers the exposed surfaces of objects. Thus, if
some portion of a wall is obscured by a cabinet, that wall area does not see direct heat input from the
room and thus should not enter into the calculation. Rather, the exposed surfaces of the cabinet
would absorb the heat. We estimate that 40% of the wall space and 15% of the floor space within a

44
 ES 96 Spring 2002

faculty office is obscured by desks, tables, cabinets, and other such items. Taking this into account,
we estimate Cr to be in the range of 4000-5000 BTU/(lb-°F). Similar calculations for a laboratory
such as the Harvard Robotics Laboratory (room 338) yield an estimated value of 6000-7000 BTU/(lb-
°F).

6.2.3 Temperature Invariant Heat Sources

As mentioned above, there are some systems in a room that will always add heat regardless of
temperature. However, these heat sources may be time-dependent. For a typical room, these include
the occupants, any electronic devices, and the sunload.

Assuming we have information about the occupancy schedule of the room and the rate at which each
of these systems adds heat to the room, the total rate of heat flow into the room from these
temperature invariant sources, Qinv, is a sum of the rates of each source.

Qinv (t) = Qpeople (t) + Qelec (t) + Qsun (t) (Eq. 6.2)

Again, we use Q to represent rates of heat gain, with units of [heat / time].
†
In practice, Qpeople is easy to estimate if one knows the occupancy schedule of the room. The energy
output of a person sitting at a desk working is approximately 400 BTU / hr, and was verified by many
independent sources. Qelec is easy to measure with wattmeters. In general, lighting provides the
majority of electrically supplied heat to a room, with four 32W fluorescent bulbs releasing heat about
as fast as one computer and CRT monitor. For reference, most of the rooms in Maxwell Dworkin
have at least twelve 32W bulbs.

Qsun is the hardest of these sources to predict because it depends on the weather. Using weather data
in part from weather data from NOAA Weather station at Tufts University in Medford, MA, and
confirmed by equations, we found that Qsun was the most significant of these temperature invariant
sources for rooms with a window facing the sun on a cloudless day. Q sun could be as high as 4000
BTU/hr at noon during January or 8000 BTU/hr at noon in the summer. These are maximum
values—depending on the hour of the day, the outside weather, and obstructions outside the window
from shading effects, we expect Qsun will be lower.

45
 ES 96 Spring 2002

6.2.4 Heat Exchanges

There are also heat exchanges between the system and its environment. One example is the rate of
heat transfer through the windows and the wall. Other exchanges may come from conditioned air
blowing into the room or radiation to the outside. In each of these cases, there is a net gain or loss of
heat that is dependent on the difference in temperatures between the external environment and the
interior of the room. As mentioned above, we ignore effects such as stratification here (measured to
be only ~0.5 oF/ft) and assume a constant temperature throughout the room—the thermostat
temperature.

The rate of heat gain due to conduction and convection through an exterior surface of the room
depends on the R-factor of that surface, Rs, the area through which the heat is conducted, As, and the
difference in temperature between the room and the external environment (Text - Tr). The total rate of
heat gain to the system through these surfaces, Qwall, is just the sum of the rates through each surface:

As (Eq. 6.3)
Qwall (t ) = Â (Text (t ) - Tr (t ) )
surfaces Rs

For one of the offices on the south wall of Maxwell Dworkin, Qwall is about 30 BTU/oF-hr times the
temperature differential. This is a low number that suggests Maxwell Dworkin is a well-insulated
building. A temperature differential of more than 13oF is required for the rate of heat passively lost
through the windows and walls to compensate for the heat supplied by a single person. Note also that
more heat is lost for greater temperature differentials. This suggests that more heat loss occurs at
higher room temperatures (Tr), a finding which could be exploited to reduce cooling costs. Our
model will demonstrate significant savings from increasing the set point temperature.

The rate of heat gain due to air changes in the room depends on the mass of air being exchanged and
the difference in temperature between the air entering the room and the air leaving the room. The rate
of heat gain is

dV (t) (Eq. 6.4)

Qair (t) = r C p,air (Tair (t) - Tr (t))
dt
where r is the density of air (0.08 lbs/ft3), dV(t)/dt is the rate at which air enters the room (measured
in units of [volume / time]), Cp,air is the specific heat of air, and T air is the temperature of the air
†

† 46
 ES 96 Spring 2002

entering the room. As will become clear from other sections in this report, Maxwell Dworkin is an
over ventilated building. The Massachusetts Building Code requires a minimum of dV(t)/dt = 20 cfm
per person, but in practice Maxwell Dworkin provides far more.

We have already considered solar radiation as a temperature invariant heat source (as part of Qinv).
Radiation energy from the room to the environment is small and compensated for in the
manufacturer's R-value for the window.

6.2.5 Active Temperature Regulation

Many of the rooms considered have a fan coil unit with heating and cooling capability and a
thermostat. Based on research on the operation of the fan coil units, we introduce the following
model for fan coil unit operation.

The fan coil unit has only two inputs: T r, the room temperature, and u t, a thermostat set-point
temperature at time t. We assume the thermostat also has some tolerance associated with it, in which
it will not modify the temperature—the dead band. Mathematically, we suppose there is some
temperature differential Tl such that the fan coil unit will heat the room if Tr < u t – Tl, but otherwise it
will not. Likewise, suppose there is some differential Th such that the fan coil unit will cool the room
if Tr > ut + Th, but otherwise it will not. Tl and Th could be the same value or different tolerances.

Assume that the fan coil unit is perfect at keeping the room temperature in the range ut - Tl < Tr < u t +
Th. We suppose that it does this by opening and closing binary heating and cooling valves as needed,
but mathematically we will write this as follows. The fan coil unit supplies heat to the room at the
rate QFCU(ut) such that the temperature in the room remains in the range ut - Tl < T r < u t + T h. As we
will see shortly, we will ultimately work with a discrete form of this model in which time is measured
in hours. The assumption that the fan coil unit controls the temperature perfectly as described above
is more accurate if the fan coil has an hour within which to reach its temperature.

6.2.6 Putting it all together

We have written rates and equations for the heat inputs and outputs to the system. The net heat gain
to the entire system is just the sum of these rates:

47
 ES 96 Spring 2002

Qr (t ) = Qinv (t ) + Qwall (t ) + Qair (t ) + QFCU (u t )

As
= Qinv (t ) + Â (Text (t ) - Tr (t ) )+ r dV (t ) C p ,air (Tair (t ) - Tr (t ) )+ QFCU (ut )
surfaces Rs dt

Ê A dV (t ) ˆ
(Eq. 6.5)
= Qsum (t ) - ÁÁ  s + r C p ,air ˜˜Tr (t ) + QFCU (u t )
Ë surfaces Rs dt ¯
where Qsum is

As dV (t )
Qsum (t ) = Qinv (t ) + Â Text (t ) + r C p ,air Tair (t )
surfaces Rs dt

With equation 6.1, we can write the differential equation governing the temperature in the room:

dTr (t ) Ê A dV (t ) ˆ
Cr = Qsum (t ) - ÁÁ  s + r C p ,air ˜˜Tr (t ) + QFCU (u t ) (Eq. 6.6)
dt Ë surfaces Rs dt ¯

Because we work with discrete hourly data, it is helpful to rewrite this equation as a difference
equation.

Tr [n + 1] = a [n]Tr [n] + 1 hr (Qsum [n] + QFCU [u n ]) (Eq. 6.7)

Cr

where

Ê A ˆ
Á C r - Â s - rV [n]C p ,air ˜
Á ˜
surfaces Rs
a [ n] = Ë ¯
Cr

and V[n] is the volume of conditioned air that enters the room between the hours n and n + 1.

This is our model for the temperature in a room.

6.2.7 Predicting the Status Quo

To demonstrate the effectiveness of the model, we conducted a five-day prediction of temperature

from April 5, 2002 to April 10, 2002 for the robotics lab in Maxwell Dworkin room 338. Small data
acquisition units were used to measure and record the temperature at distributed locations as well as

48
 ES 96 Spring 2002

light levels and fan coil unit activity. The external temperature data came from a commercial weather
station at Tufts University in Medford, MA, within five miles of Maxwell Dworkin11.

The thermostat was set at a level far below the setting allowed by the APOGEE system. We assumed
a tolerance of Tl = T h = 2 oF and chose ut to match the observed daytime data. ut remained constant
until the final full day of the trial (April 9), where it appears that ut increased by a degree.

FIGURE 6.2. Prediction of the temperature for MD338 and actual temperature
data over a five day period.

Figure 6.2 shows the correlation between the predicted temperature of the room and the actual
temperature, as observed by a data acquisition unit next to the thermostat. The correlation is fairly
good. The discrepancies between the models are probably due to errors in the estimates of when the
room was occupied. During the night of April 9 - 10, the data acquisiton units recorded a significant
temperature drop that was not caused by the fan coil units. Preliminary explanations are a significant
air change in the room, perhaps from opening the window. The model did not include such

11
Weather data from NOAA Weather station at Tufts University, Medford, MA.

49
 ES 96 Spring 2002

occurrences, and therefore we see a significant discrepancy between the model and the actual
temperature for that time.

6.3 Findings

6.3.1 Application of the Model

Armed with a verified temperature model, we can now predict how changes in Maxwell Dworkin’s
control strategy will affect the temperature characteristics of a room. Figure 6.3 shows a one-day
simulation run for a graduate student office (Room 342) in Maxwell Dworkin with June 1st weather
data. The office has one exterior wall with two windows and two FCUs. Since the windows face
north, the sunload is negligible. Investigation of the actual room showed that there were four
computers and four CRT monitors in the room and twenty-four 32W fluorescent lamps (2620
BTU/hr) hanging from the ceiling. From observations of student behavior, occupancy was estimated
to be four people between the hours of 9am and 8pm, with one person remaining until 11pm. When
the room was occupied, we assumed that all lights were on, and that one computer and monitor
combination was active for each person present (480 BTU/hr for each computer-monitor pair). All
unused computers were assumed to be in sleep mode (at 120 BTU/hr) and all unused monitors were
assumed to be off. Thus, between the hours of 11pm and 7am the following day, the only source of
heat would be from the four computers in sleep mode. All values for computer and monitor power
consumption are from the Harvard Green Campus Initiative12.

12
Harvard Green Campus Initiative, http://www.greencampus.harvard.edu/green_projects/cerp/facts.html

50
 ES 96 Spring 2002

FIGURE 6.3 Simulation of graduate student office (i.e. Maxwell Dworkin

room 342) on June 1 with constant thermostat setpoint of 72ºF and dead band
of 2ºF. Room has two north-facing windows and two FCUs.

In the simulation, there is a two-degree dead band, making the constraint between 71ºF and 73ºF.
During the night, temperature remains essentially constant at 76ºF, indicating that the heat emitted by
the four computers in sleep mode nearly balances the passive loss through the windows. At 7am, the
APOGEE system becomes active, and the FCUs immediately begin circulating chilled water to bring
the temperature within the control band. From this point until 7pm, the FCUs maintain the
temperature at the top of the control range, since such a practice minimizes the amount of cooling
required—and thus cost incurred. At 7pm, the APOGEE system deactivates, and the FCUs shut off.
The temperature then begins to rise, first from the four occupants present until 8pm, and then from the
single occupant remaining until 11pm. The temperature rise ceases when this late-worker departs,
remaining steady at 76ºF until 7am the following morning.

The most striking feature of this narrow-band constraint is the constant need for cooling during the
day. Active heating is almost never required, even in the colder months of the year, as was evidenced
by a simulation done with this model for an entire year’s worth of data. Cooling comes at a higher

51
 ES 96 Spring 2002

cost than heating ($30.21 per MMBTU cooling versus $17.97 per MMBTU heating)13. Indeed, there
is free heating from the sunload and human heat. Furthermore, both steam and chilled water are used
during some months of the year, implying that there may be both heating and cooling of rooms on a
single day14. Such a scenario is easily imagined, for on a very cold evening, the passive heat loss
through a window may outweigh the input from the computers in sleep mode. Thus the temperature
of the room would drop during the night, perhaps falling below the lower limit of the narrow band
constraint. An example of this scenario is shown in Figure 6.4, which represents a simulated office
during a day in January. In this case, at 7am the subsequent morning, the FCUs would have to
circulate hot water to warm the room. However, after this occurs, people begin to arrive at work,
turning on lights and equipment and thus increasing the heat input to the room. At this point, the heat
flowing into the room may exceed the passive loss, causing the room temperature to rise until it
exceeds the upper limit of the control band, initiating cooling. Clearly, such a costly scenario should
be avoided.

13
FAS Planning numbers for Fiscal Year 2003. Courtesy of the Office of Physical Resources.
14
Steam and Chilled Water billing data, University Operations Services

52
 ES 96 Spring 2002

FIGURE 6.4. Simulation of an office for a day in January demonstrating

the risks of heating and cooling in the same day. The first bump in the
conditioning energy between hours 7 and 8 represents hot water heating
of the room, while the subsequent steady rise represents chilled water
cooling. These risks are associated with a narrow band constraint, here
shown as 2ºF for emphasis.

6.3.2 Optimizing the Office Controls

Our temperature model allows us to determine an optimal strategy for regulating a room’s
temperature in Maxwell Dworkin by formulating it as a control problem. For our purposes, we will
define the optimal strategy for temperature regulation as the strategy that minimizes the money spent
in heating and cooling at the FCU. This objective can be expressed as:
•
Minimize over u: Â f (Q FCU [uk ])g k (Eq. 6.8)
k =1

where f(QFCU[uk]) is a function that converts the amount of heat added or removed by the fan coil unit
into a cost and g Π(0,1) is a discount factor that weights immediate costs higher than costs in the
future. We define f(QFCU[uk]) from the costs for heating and cooling given above and in the FAS
Planning number for Fiscal Year 2003.

53
 ES 96 Spring 2002

We can minimize this sum subject to constraints defined by people’s comfort levels. For example,
we can make the constraint that the room temperature must always be between 65oF and 85oF, i.e. 65
£ T r £ 85. Furthermore, this temperature range may vary with time, relaxing to this range at night,
and becoming more strict when the room is occupied. Another constraint could be that the
temperature can only change so much from hour to hour, i.e. Tr[n] – T d[n] < T r[n + 1] < Tr[n] + T d[n].
For example, a value of Td[n] = 1 for some n represents the constraint that the temperature at time n +
1 must be within one degree of the temperature an hour earlier. This constraint would protect against
rapid temperature swings that some might find uncomfortable. These are just two plausible
constraints—this problem can be solved under any feasible constraint.

The formulation of our problem is:

•
Minimize: Â f (Q FCU [uk ])g k
k =1

1
State Evolution Equation: Tr [n + 1] = a [n]Tr [n] + (Qsum [n] + QFCU [un ])
Cr
Subject to: Tmin[n] £ Tr[n] £ Tmax[n]
Tr[n] – Td[n] < Tr[n + 1] < Tr[n] + Td[n]

One can solve this problem in a number of ways. We solved it using reinforcement learning, but this
is a simple dynamic programming problem and can be solved using many other methods. The
solutions to this problem are not expressible in closed form and require a computer to solve; however,
we quickly note that the solution of this problem is almost intuitive. The less we have to regulate the
temperature, the more money we save.

54
 ES 96 Spring 2002

FIGURE 6.5 Winter and summer comfort zones for people

in typical winter and summer clothing during primarily
sedentary activity. From 2001 ASHRAE Handbook.15

Because our fan coil unit operation overlaps fairly well with when Maxwell Dworkin is occupied, an
obvious solution would be to raise the thermostat setpoint and widen the dead band beyond the four
degrees currently used. To determine how much widening should occur without compromising
comfort, we consulted the ASHRAE comfort zones, depicted in Figure 6.5. These zones are plotted
on what is known as a psychrometric chart. Such charts allow us to characterize a volume of air
based on temperature and relative humidity. The hatched regions on the chart depict the comfort
zones as determined by submitting test subjects to air of varying temperatures and humidity levels
and recording their comfort level. From the chart, we approximated the ideal temperature ranges to
be between 69ºF and 75ºF in the winter and 74ºF and 80ºF in the summer. Returning to our June 1
graduate student office simulation, we included this new information, widening our control band to
encompass the full summer comfort zone. Additionally, we utilized a floating setpoint strategy.

55
 ES 96 Spring 2002

Instead of keeping the setpoint at a fixed temperature, we allowed it to self-adjust throughout the day.
The resulting plot, presented in Figure 6.6, at first glance looks very similar to the narrow band
constraint strategy seen in Figure 6.3. However, the important difference is that the temperature of
the room is at all times higher than with the narrow band constraint. Between 7am and 7pm, we again
observe that the FCUs cool the room to remain at the top of the temperature range—80ºF. With this
widened control band and floating setpoint, cooling cost is greatly reduced. Running this simulation
with weather data for the entire first week of June, savings of 18% over the narrow band strategy
were possible.

FIGURE 6.6 Simulation of same graduate student office as in Figure 3 on June 1

with floating thermostat setpoint and control band broadened to encompass
summer comfort zone of 74ºF to 80ºF.

However, this example assumed a two-degree dead band, whereas Maxwell Dworkin actually
operates with a four-degree band. We incorporated this factor into the aforementioned simulation run
with an entire year’s worth of data. We assumed the same room parameters as in the simulations

15
2001 ASHRAE Handbook

56
 ES 96 Spring 2002

already presented. It is here that we found the amount of heating done by the FCUs to be negligible.
Comparing the narrow band and optimal strategies via this simulated year, we found that savings up
to 34% were possible. Realizing that the room simulated may not be entirely typical of the spaces in
Maxwell Dworkin and that other changes to building control would affect the need for FCU cooling,
we selected a more conservative estimate of 20% savings. Realizing that this savings comes almost
entirely from chilled water use in the FCUs, we calculated a cost savings of approximately $14,000
per year, or an equivalence of 30 acres of US commercial forest.

6.4 Recommendations

From the observations and simulations described above, we feel that an optimal control strategy
utilizing either a widened control band or an intelligently floating setpoint should be implemented in
Maxwell Dworkin. These changes would require a one-time cost to modify the APOGEE control
system’s programming. Indeed, widening the dead band should be an almost trivial task. However,
some people may regard 80ºF as uncomfortably warm during the summer, despite the ASHRAE
suggestions. If this is the case, the control band could be slightly narrowed or shifted down to
encompass a slightly lower temperatures. While these would reduce the amount of savings that could
be reaped, they would ensure that the occupants of the building remained comfortable. Ideally, the
temperature control of a few volunteers’ rooms would be switched to this new strategy as a case study
before employing it building-wide. Additional simulations could also be run with the model to assess
the effect on savings if the band were narrowed or shifted. Regardless of the strategy chosen, our
model clearly demonstrates two trends that make intuitive sense:
1. Wider control bands are cheaper to maintain than narrower control bands
2. Higher building temperatures are cheaper to maintain year round than lower temperatures.

These two trends are direct consequences of operating a well-insulated building. The heat supplied
by the building’s electrical load alone is enough to require cooling on many days of the year. In
choosing a strategy that follows either of these two trends, there is potential for significant
savings—both in dollars and in adverse environmental impact—at comparatively little cost.

57
ES 96 Spring 2002

58
 ES 96 Spring 2002

7.0 COOLING IN MAXWELL DWORKIN

59
ES 96 Spring 2002

60
 ES 96 Spring 2002

7.1 Introduction

Maxwell Dworkin is such a well-insulated building, and, as a result, the majority of the heat that
enters the building via steam, electricity, sunload, and other means must be actively removed through
the building’s chilled water system. During our investigation of Maxwell Dworkin, we discovered
various inefficiencies in the way that this cooling is being controlled on a daily, weekly, and seasonal
basis and in the management and operation of the control system in general. The results of these
inefficiencies are surprising trends in chilled water usage, such as the significant use during the winter
months. In order to reduce chilled water consumption in Maxwell Dworkin, we recommend a
combination of certain changes to the scheduling of cooling within the building and the use of
alternative forms of cooling.

7.1.1 Chilled Water System in Maxwell Dworkin

Chilled water is produced in a facility in the basement of the Science Center using large machines
called chillers, distributed to the various buildings on campus via buried insulated pipes, and then
returned to the Science Center for re-cooling. The chilled water facility attempts to maintain a supply
water temperature of 45ºF and requests that each building return its used chilled water at
approximately 60ºF, in order to achieve maximum efficiency in the cooling process.

Chilled water is measured in a unit of energy called a ton-day, which is equivalent to approximately
288,000 BTUs. Each ton-day of chilled water is billed by the university at $8.70, based on planning
figures for the fiscal year 200316. Chilled water is metered by recording both the flow rate of the
water, in thousands of gallons per minute, to each building and the temperature difference between
the water supplied and the water returned. These two values are used to calculate the ton-day unit for
billing purposes in two-hour intervals.

16
From Eugene Arcand, FAS Operations.

61
 ES 96 Spring 2002

Once the chilled water reaches Maxwell Dworkin, it enters the building through a mechanical room in
the basement, where it is pumped throughout the building to provide cooling. It is distributed to the
five Air Handling Units, all of the fan coil units on the top three floors, and a stand-alone cooling unit
in the DEAS computer server room in the basement. At these various locations the water may absorb
heat from the air, if cooling is necessary, and experience a rise in temperature. It is then returned to
the mechanical room where it entered the building and may either be returned to the Science Center
facility or re-circulated within the building to provide more cooling. Currently, the APOGEE control
system, which controls the flow of chilled water through the manipulation of a set of valves, is set to
re-circulate the water when its temperature is below 60ºF. Once the water temperature reaches the
setpoint of 60ºF, it is returned to the Science Center facility, in accordance with the facility’s
requested return water temperature.

A very basic diagram of the chilled water system within Maxwell Dworkin is shown in Figure 7.1:

FIGURE 7.1 Diagram of the Chilled Water System in Maxwell Dworkin

62
 ES 96 Spring 2002

In this diagram, the box marked cooling load represents the various devices within the building that
use chilled water to provide cooling. The valves shown are a simplified representation of the valves
controlled by the APOGEE system in order to maintain a set return temperature.

While inside Maxwell Dworkin, chilled water is used to provide two main of types of cooling:
comfort cooling and computer load cooling. Comfort cooling is provided to all areas within the
building that are occupied on a regular basis, such as offices, research spaces, classrooms, lecture
halls, and common spaces. On the ground floor and in the basement, this type of cooling is provided
strictly through Air Handling Units, while on the upper three floors, it is accomplished via a
combination of the Air Handling Units and the fan coil units within each office, research space,
classroom, and common area. The amount of chilled water required to provide adequate comfort
cooling in Maxwell Dworkin is difficult to estimate because it depends on a variety of factors,
including occupancy levels, electrical usage, and outside air temperature. Currently, the APOGEE
system is scheduled to provide comfort cooling 7 days a week from 7 a.m. to 7 p.m.

Computer load cooling is provided for areas within Maxwell Dworkin that contain sensitive electrical
equipment, which requires precise control of environmental conditions. There are four such areas in
the building: the DEAS computer server and telephone equipment rooms in the basement and two
EECS mechanical rooms on the first and second floors. By visiting these areas, we were able to
estimate the total amount of electrical power consumed by the various pieces of equipment to be
approximately 20 kilowatts, and thus estimate the amount of cooling necessary. The computer server
room has a stand-alone power supply unit, which can provide data on total electrical usage within the
room, the telephone equipment room contains very little equipment, and the two EECS rooms contain
similar numbers and types of units to the server room.

7.2 Method of Investigation

Our research of the chilled water system in Maxwell Dworkin proceeded along two main avenues: the
acquisition of chilled water usage statistics for the entire building and the investigation of the control
system and mechanical units that provide cooling within the building. Using the data thus acquired,
we were able to identify potential areas for improvement in the operation of the chilled water system.

63
 ES 96 Spring 2002

We were able to contact Ms. Susyrati Bunanta, P.E., at the chilled water facility in the Science Center
and obtain records of Maxwell Dworkin’s chilled water use in two-hour intervals for an initial two-
week period. Throughout the term, we received frequent updates of these records and eventually
compiled the building’s chilled water usage statistics for a period from March 7, 2002 to April 30,
2002. By examining these statistics and the various trends inherent within them, we gained an insight
into the cooling schedules and relative hourly and daily cooling levels of Maxwell Dworkin. Several
of the more interesting trends are presented in the next section.

After discovering some interesting results in the chilled water usage trends for Maxwell Dworkin, we
attempted to determine their sources by examining the way cooling is controlled and provided for in
the building. We were able to closely examine and study the building’s Air Handling Units and fan
coil units, which provide most of the cooling, in order to determine how they operate. We also visited
each of the four rooms that are provided with 24-hour computer load cooling, in order to determine
the amount of cooling they require and the means by which they are cooled.

Finally, we examined in detail the way in which the APOGEE control system schedules and manages
cooling and chilled water use in Maxwell Dworkin, with the help of Greg Kousidis of Facilities
Maintenance Operations. We explored the scheduling capabilities of the control system, as well as
the current cooling schedules for various groups of rooms in the building. We were able to verify the
12 hours a day, 7 days a week cooling scheme for the top three floors. During our investigation of the
abnormally high level of night-time cooling, we obtained a list of the rooms in Maxwell Dworkin that
were being cooled on an alternate schedule and were able to identify several situations in which
cooling was being done unnecessarily. We also used APOGEE’s data-trending capabilities to
monitor hourly chilled water usage and the operation of the valve system, which controls the flow of
chilled water back to the Science Center facility.

7.3 Findings

Close observation of the operation of a system is an important step in attempting to determine energy
conserving steps that could be taken. Often, scrutiny of current operating conditions reveals
inefficiencies and mistakes that are easily corrected. As such, we decided to trend the chilled water
usage in Maxwell Dworkin over a month long period in order to spot any anomalous behavior. Every

64
 ES 96 Spring 2002

two hours, the number of ton-days of cooling the building was using was metered at the chilled
water’s exit point from the building. This time increment allowed us to examine trends in chilled
water use on a daily and weekly basis. Furthermore, data available from University Operations
Services allowed us to observe the trends in monthly use. This section outlines these trends.

7.3.1 Daily Use of Chilled Water

Our discussions with Greg Kousidis and his presentation of the APOGEE system helped us
understand that the building was cooled from 7 a.m. until 7 p.m. The trended data found in Figure 7.2
reveals that this is, in fact, the case:

Chilled Water Usage in 2-hour Intervals for Maxwell Dworkin

Averaged from March 8 to April 2, 2002

3.5

2.5
Ton-days

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20 22
Hours past midnight

FIGURE 7.2 Chilled Water Usage in 2-hour Intervals for Maxwell Dworkin
Averaged from March 8 to April 2, 2002

Approximately 1 ton of cooling is used from 7 pm to 7 am, when the building is “off.” When cooling
comes on at seven in the morning, however, the cooling load trebles. Two conclusions could be
drawn from these trends: first, that some parts of the buildings are being cooled twenty-four hours a
day; second, that the building was warm enough throughout the month of march that it actually
needed to be cooled in the mornings. That is, the building was actually being heated at night.

65
 ES 96 Spring 2002

7.3.2 Weekly Use of Chilled Water

Further analysis of chilled water use on a weekly basis revealed, most surprisingly, that the cooling
load was constant throughout the week, as demonstrated in Figure 7.3:

Daily Chilled Water Consumption in Maxwell Dworkin

Averaged from March 8 to April 2, 2002

35

30

25
Ton-days

20

15

10

0
Monday Tuesday Wednesday Thursday Friday Saturday Sunday
Day of the Week

FIGURE 7.3 Daily Chilled Water Consumption in Maxwell Dworking

Averaged from March 8 to April 2, 2002

This graph presents the average daily cooling load for the monitored period.) That is, the building was
drawing as much cooling on Saturday and Sunday as it was Monday through Friday. This
contradicted information received from Greg Kousidis and Jay Phillips, who believed the building
was cooling at its “night-time” load during the weekend. Further discussion with Greg Kousidis,
however, and scrutiny of the APOGEE System revealed that the building was, in fact, set to cool on
the weekends.

66
 ES 96 Spring 2002

7.3.3 Monthly Use of Chilled Water

Using information from the University Operations Services website, monthly chilled water use data
was acquired for the three years of the building’s existence. These figures were then averaged out on
a per monthly basis and are presented in Figure 7.4:

Monthly Chilled Water Usage for Maxwell Dworkin

2000

1800

1600

1400

1200
Ton-Days

1000

800

600

400

200

0
r

st
r

er

ay
y

ril

ne
ch

ly
be

be
be

ar

ar

Ju

gu
Ap
ob

M
ar

Ju
nu

u
em

em
m

Au
br

M
ct
e

Ja
O

ov

ec
pt

Fe
Se

Month

FIGURE 7.4 Monthly Chilled Water Usage for Maxwell Dworkin

As could be expected, the bulk of the building’s cooling occurs during the months of July, August,
and September. These months average over 1600 ton-days each and constitute over 40% of the
building’s total cooling cost. More surprisingly, however, analysis of chilled water use through these
months revealed that the building is being cooled during the winter. Significant amounts of heat are
being mechanically removed from the building when the outside air temperature is far below the inner
air temperature set point ,at great cost to the building operations.

67
 ES 96 Spring 2002

7.3.4 24-Hour Cooling

As was shown by the hourly chilled water trends for Maxwell Dworkin presented in the previous
section, there is a roughly constant nighttime cooling load in the building, which is equivalent to an
approximately 45-kilowatt electrical load. However, as previously mentioned, the only four rooms
that are intended to be cooled on a round-the-clock basis—the computer server room, telephone
equipment room, and two mechanical rooms—account for only about 20 kilowatts of electrical power
use. Therefore, there must be some other space within Maxwell Dworkin that is being cooled on a
24-hour schedule.

Although all other spaces within the building are set to be cooled only from 7 a.m. to 7 p.m., the
APOGEE control system as the capability for programmed “exceptions”, by which any space or
group of spaces within Maxwell Dworkin may be cooled on an alternative schedule at the request of
the system operator. Therefore, should an occupant of any space in the building request extra hours
of cooling, a permanent change may be made to the cooling schedule for that space.

Unfortunately, many times these changes are made on an informal basis by the system operator and
then forgotten. This may result in various spaces being cooled unnecessarily when they are not
occupied. Furthermore, these “exceptions” are not readily apparent to the system operator during
normal operation and can only be found if they are actively searched out.

During the last week of April, system operator Greg Kousidis used the APOGEE system to create a
list of all of the spaces within Maxwell Dworkin that were being cooled on an alternative schedule.
The list showed the following 13 spaces in Maxwell Dworkin were being cooled 24 hours a day in
addition to the four rooms previously mentioned17:
• Offices and research spaces: 235, 307, 311, 321, 323, 329, 333, 334, 340
• Common areas: L-22, L-31, L-32
• IT Department office: G-107

At the time of the report’s creation, Mr. Kousidis was able to cancel the schedule overrides for each
of these spaces and return them to the normal cooling schedule of 7 a.m. to 7 p.m. It is unknown how

17
Email from Greg Kousidis, April 24, 2002.

68
 ES 96 Spring 2002

long these spaces had been excepted from the normal cooling scheme, and it is difficult to estimate
the amount of wasted cooling due to varying loads within the spaces.

However, even after these spaces had been returned to the normal cooling schedule, no significant
drop in nighttime chilled water consumption occurred. Figure 7.5 shows the average chilled water
usage in Maxwell Dworkin in 2-hour intervals for a period from April 24 to April 29, 2002:

Chilled Water Usage in 2-hour Intervals for Maxwell Dworkin

Averaged from April 24 to April 29, 2002
4

3.5

2.5
Ton-days

1.5

0.5

0
0 2 4 6 8 10 12 14 16 18 20 22

Hours since Midnight

FIGURE 7.5 Chilled Water Usage in 2-hour Intervals for Maxwell Dworking
Averaged from April 24 to April 29, 2002

Thus, at present we are unable to account for roughly 50% of the nighttime chilled water use in
Maxwell Dworkin. It is possible that the discovery of the cause of this discrepancy could lead to a
reduction in chilled water consumption.

69
 ES 96 Spring 2002

7.4 Recommendations

After examining the data acquired on the chilled water system in Maxwell Dworkin, we offer the
following proposals for ways in which the building could more efficiently be cooled in order to
reduce chilled water consumption. These suggestions can be broken into two main categories:
scheduling and alternative forms of cooling. Simulations based on the office simulation model
presented earlier are used to model the effects of these suggestions on building temperature and total
cooling load. Since many of these proposals involve only manipulation of the APOGEE control
system, they involve little or no initial cost. Savings estimates are provided with each suggestion for
which they are applicable.

7.4.1 More Efficient Operation of the APOGEE Control System

In order to prevent situations in which spaces are unnecessarily excepted from the building’s normal
cooling schedule, we recommend a more strict maintenance of the building’s cooling schedule via the
APOGEE system. There are various alternatives to the present scheme for managing the scheduling
of cooling that could prevent waste of chilled water.

First, the APOGEE control system possesses the capability for pre-programmed one-time
“exceptions”, by which any space or group of spaces within Maxwell Dworkin may be cooled on an
alternative schedule for a set period of time and then returned to the normal schedule. Thus, if a
space is expected to be occupied until late hours for only a day or a week, the cooling schedule need
not be changed on a permanent basis to accommodate.

Another means of more efficiently managing the cooling control system within Maxwell Dworkin is
to formalize the process by which alternative cooling schemes may be requested and arranged. Each
thermostat in the building is equipped with a three-hour override, which may be used by occupants to
provide cooling within a space for a three-hour period at any time of the day. Therefore, if the
occupant of a space occasionally stays past the normal 7 p.m. cooling shutdown, he or she may
request additional cooling on a day-to-day basis, rather than requesting that the spaces’ cooling
scheme be changed permanently.

70
 ES 96 Spring 2002

Finally, a more organized system for processing and recording changes to the cooling scheme must be
arranged. Clear records should be kept of any changes that are made via the APOGEE system, and
routine checks of the system for unusual behavior should be made. In this way, changes to the
cooling scheme could be monitored and subsequently updated if they are found to be unnecessary.

Closer monitoring of the APOGEE system can be accomplished with minimal additional effort over
present system operation. Therefore, all of the suggestions proposed here can be included within the
duties of personnel already on staff and thus require no additional staff members.

7.4.2 Elimination of Weekend Cooling

The nearly constant daily chilled water consumption throughout the week in Maxwell Dworkin
results from the 7 day a week cooling schedule that is currently used. However, because many of the
spaces in the building are not occupied on the weekends, cooling may only be necessary 5 days a
week. An unoccupied space has a reduced heating load, as a result of lower occupancy levels and
electrical usage, and its temperature may also vary over a wider range of temperatures without
consequence.

Using an adaptation of the office simulation model presented earlier, we modeled the effects of not
cooling an unoccupied faculty office for a typical weekend in July. We assumed a high outside air
temperature of 80ºF on a perfectly sunny day with the only electrical heat input coming from a
computer in sleep mode. Under these assumptions, the following temperature profile (Figure 7.6)
would result:

71
 ES 96 Spring 2002

FIGURE 7.6 Room temperature as a function of time on weekends without cooling or

air purges in July

As this graph shows, the temperature inside the office may exceed 120ºF on Sunday afternoon,
which, although safe as long as the office is unoccupied, may be considered undesirable for various
other reasons.

However, should nighttime air purges, which are discussed in the next section, also be implemented,
the temperature effects of eliminating weekend cooling can be significantly reduced. The following
figure shows the temperature profile of an office under the same conditions (Figure 7.7), if the room
temperature is lowered to 64ºF at 3 a.m. each morning using the proposed system of purges:

72
 ES 96 Spring 2002

FIGURE 7.7 Room temperature as a function of time on weekends with no cooling but
air purges

Using this system, the temperature inside the office peaks at approximately 90ºF on Sunday
afternoon, which is a much more acceptable level than the one that results from the previous scenario.
Therefore, we conclude that if a system of nighttime air purges is instituted, then cooling may be
effectively eliminated on the weekends even in the summer with only moderate effects on the
temperature of unoccupied spaces.

For any spaces within the building that are occupied on weekends, the three-hour thermostat override,
which was described in the previous section, may be used to provide active temperature control.
Thus, weekend cooling can be provided on a room-by-room basis, and unnecessary cooling of
unoccupied spaces can be eliminated.

73
 ES 96 Spring 2002

The main effect of eliminating weekend cooling in Maxwell Dworkin would be to reduce chilled
water consumption on the weekends to near current nighttime levels. We can estimate potential
savings by assuming that chilled water usage on each weekend of the year can be reduced to a
constant 24-hour level plus a reduced day-time level, which results from cooling that may be needed
in occupied spaces.

To produce the savings estimate presented below, we assumed that the pattern of nearly constant
daily chilled water consumption that is evident in March and April is consistent throughout the year.
Therefore, we can estimate the current amount of cooling provided on a given weekend day by
dividing a monthly average by the number of days in the month. We also made two relatively safe
assumptions in computing the amount of cooling that is necessary for a given weekend day. First, we
assumed that the current nighttime level, of which 50% is unaccounted for, is constant throughout the
year. We also included a safety margin by assuming that 20% of the daytime cooling currently used
is required for occupied spaces. The Table 7.1 presents the savings estimates under these
assumptions:

TABLE 7.1
Savings Estimates for Eliminating Weekend Cooling
Weekend Cooling (ton-days/day) Savings (ton-days)
Month Current Necessary Daily Monthly
July 59.32 23.06 36.26 290.06
August 59.92 23.18 36.74 293.88
September 54.68 22.14 32.55 260.37
October 36.63 18.53 18.11 144.86
November 34.67 18.13 16.53 132.27
December 24.73 16.15 8.58 68.68
January 22.58 15.72 6.86 54.92
February 22.26 15.65 6.61 52.88
March 19.62 15.12 4.50 35.99
April 26.28 16.46 9.83 78.61
May 35.05 18.21 16.84 134.71
June 33.25 17.85 15.40 123.20

74
 ES 96 Spring 2002

By totaling these savings over an entire year, we arrive at the following estimates for annual savings
(Table 7.2):

TABLE 7.2
Annual Savings by Eliminating Weekend Cooling
Ton-days 1670
Dollars $14,529
Acres of US forest 29.8

This suggestion is very simple to implement in its current form and requires only a few clicks of the
mouse in the APOGEE control system. If the complete elimination of weekend is found not to be
desirable, a more complex system which takes into account such factors as outside air temperature
could be implemented by adding code to the current APOGEE control.

7.4.3 Summer Night-Time Air Purges

In Maxwell Dworkin, the peak cooling load occurs not surprisingly during the summer months of
July, August, and September, which account for roughly 40% of yearly chilled water consumption.
Therefore, a reduction in the cooling load during these months would have a significant effect on
overall chilled water usage.

To this end, we propose a system of nighttime air purges in Maxwell Dworkin, which would take
advantage of cooler nighttime air to reduce the cooling load during the day. During the summer
months, the outside air temperature usually reaches its low between 2 a.m. and 4 a.m., when the
building is largely unoccupied, and so there exists a means of lowering the inside temperature at the
start of the day. This would result in delaying the onset of active cooling in the building by a few
hours in the morning as the building’s temperature gradually rose to the top of the control range.

Maxwell Dworkin’s Air Handling Units require only about one hour to complete an air change of the
top three floors and thus could reduce the air temperature on these floors to near the outside air
temperature as long as the outside air temperature and humidity were favorable. Temperature data for
Boston shows that the average low temperature during July, August, and September is in the low to

75
 ES 96 Spring 2002

mid 60s, so the air on the top three floors could be cooled to nearly 65ºF each night. By determining
the viability of such a purge by comparing the outside air enthalpy to the inside air enthalpy, purges
that bring in hot or humid air could be avoided.

Using a version of the office simulation model already presented, we simulated the effects of a
nighttime air purge for a typical July day in Maxwell Dworkin. For this simulation, we assumed a
high temperature of 80ºF on a perfectly sunny day, one person occupying of a south-facing faculty
office from 9 a.m. to 6 p.m., and electrical heat provided by all of the lights in the office, a computer,
and a computer monitor. Figure 7.8 shows the temperature profile and cooling load for an office
under these conditions if no nighttime air purge occurs:

FIGURE 7.8 Room temperature as a function of time without night-time air purges in July

As this graph shows, the temperature in the office remains roughly constant throughout the night
because Maxwell Dworkin is well insulated and then rises to 78ºF at 7 a.m. At this time, active

76
 ES 96 Spring 2002

temperature control begins via the fan coil unit, and cooling starts and is needed throughout the day.
The total daily cooling load for an office under these conditions is approximately 117,000 BTUs.

Figure 7.9 represents an office under the same conditions, except with an air purge at 3 a.m., which
lowers the inside air temperature to 64ºF:

FIGURE 7.9 Room temperature as a function of time with night-time air purges in July

The temperature rises throughout the early morning, reaching a comfortable 72ºF at 9 a.m., but does
not reach the 76ºF upper control bound until approximately 10 a.m., at which time cooling is
required. Therefore, three fewer hours of cooling are necessary during the day, and the total cooling
load is only about 78,000 BTUs, a 33% decrease.

A system of such air purges each night would reduce daily chilled water consumption on weekdays
during the peak cooling months of July, August, and September. By considering only the 45 faculty
and staff offices in Maxwell Dworkin, we can estimate savings in chilled water usage using the

77
 ES 96 Spring 2002

cooling load figures from the simulations above and from similar simulations for August and
September (Table 7..3):

TABLE 7.3
Savings Estimates for Night-Time Air Purges
Daily Cooling Load Per Office (BTUs) Daily Savings (ton-days)
Month Without Purges With Purges Per Office Total
July 117000 78000 0.14 6.09
August 105000 69000 0.13 5.63
September 84000 55000 0.10 4.53

To compute total savings, we assumed that nighttime air purges could be employed on 20 weeknights
per month, which allows for exceptions, such as nights when the outside air is very humid. Under
this assumption, the annual savings are as follows (Table 7.4):

TABLE 7.4
Annual Savings for Night-time Air
Purges
Ton-days 325
Dollars $2,828
Acres of US Forest 5.8

These figures represent savings only for the 45 faculty and staff offices within Maxwell Dworkin.
Additional savings are possible in research spaces, classrooms, and common areas as well.

The proposed system of night-time air purges would not be difficult to implement, as it would require
only the addition of code to the APOGEE control system to perform the enthalpy calculations and
schedule the necessary operation of the Air Handling Units. No additional personnel or equipment
would be required.

78
 ES 96 Spring 2002

7.4.4 Ethylene Glycol Heat Exchangers

The chilled water provided by the plant in the basement of the Science Center achieves a remarkably
high Energy Efficiency Ratio of 12. Consequently, any changes to the equipment of Maxwell
Dworkin, or any future buildings, must consider the impacts of this Energy Efficiency Ratio in their
decision-making. In order to explore whether alternative methods of cooling were economically
feasible, we examined the possibility of installing an ethylene-glycol heat exchange mechanism, as
shown in Figure 7.10, that would perform some “water-side economizing.”

FIGURE 7.10 Proposed Ethylene Glycol System

As the diagram shows such a system would entail closing the return valve in the winter months and
cooling the chilled water used in the building by means of heat exchange. The “warmed” chilled
water would give off it’s heat to an ethylene glycol solution – antifreeze – which would then be
circulated through a rooftop system to give off its energy to the atmosphere.

79
 ES 96 Spring 2002

Discussions with Michael Kjelgaard, P.E., of Shooshanian Engineering revealed that “water side
economizers” had the potential to run for over 3500 hours over the course of 2001. That is, installing
a system to take advantage of colder outside air could have saved some 145 days of cooling. Mr.
Kjelgaard, however, felt that a system of rooftop exchanges would be impractical and highly
expensive, in particular because of the toxicity and corrosivity of ethylene glycol under certain
conditions.

Our discussions with Mr. Kjelgaard as well as others involved with the cooling of Harvard’s
Buildings also lead us to consider some alternative forms of cooling. In particular, our analysis led us
to research reciprocating chillers, which can achieve Energy Efficiency Ratios of 12 to 14, as
potential alternatives to the current system. Stanford University has recently installed such a system
to reduce its peak electrical demand.18 Unfortunately, we have been unable to explore the feasibility
of such an option with William Flanagan, the Associate Manager of Mechanical Utilities and
Construction for the Central Utilities Plant.

7.4.5 Steam to Hot Water Heat Exchanger

Another source of savings we studied was the steam to hot water heat exchanger, which provides hot
water used to heat the rooms. Figure 7.11 shows the operating sequence of a heat exchanger.

Figure 7.11 A typical operating sequence for a heat exchanger

18
http://www.stanford.edu/dept/news/report/news/april21/iceplant-421.html

80
 ES 96 Spring 2002

Steam enters the heat exchanger from the plant where it condenses, giving up its heat to the supply
hot water. Condensation from the steam is collected and returned to the hot water plant. The hot
water for heating is circulated throughout the rooms then returned to the heat exchanger where it can
be reheated to the set temperature. In Maxwell Dworkin, there is one heat exchanger that is used to
provide heat to all the fan coils in all the rooms, all the air handling units, and all the VAV boxes.

Chilled water savings may be possible by making changes to the heat exchanger would lower
parasitic heat loss in Maxwell Dworkin. The temperature of the hot water for heating provided by the
heat exchanger is set to be 190 degrees on the coldest days with an outdoor temperature of 0 degrees
and only drops to 170 degrees for warmer days of 60 degrees. Conventional settings for other
buildings, including many of Harvard’s own buildings, are 170 degrees for the coldest days and 120
degrees for warmer days. It is desirable to lower the temperature of the supply hot water as such
because it prevents or reduces parasitic heat loss. Parasitic heat loss occurs when hot water is passed
through the rooms, and even if it is not used, there is an unwanted transfer of heat to rooms
throughout the building. Lowering the temperature of the supply hot water would mean less parasitic
heat loss and therefore result in savings due to less chilled water being used to cool the rooms.
Savings are difficult to estimate because it is unclear just how much the reduction in parasitic heat
loss to rooms would be.

7.5 Summary

Overall, we are recommending various changes to the chilled water system in Maxwell Dworkin that
will reduce chilled water consumption by minimizing unnecessary cooling of unoccupied spaces and
by using alternative forms of cooling. These proposals include more efficient maintenance of the
cooling schedule via the APOGEE system, elimination of general weekend cooling throughout the
year, a system of summer night-time air purges, updating the heat exchanger settings, and perhaps an
enhancement of the chilled water system using ethylene glycol or a similar substance.

81
 ES 96 Spring 2002

Total annual savings, based on the savings estimates presented above when possible, are summarized in Table
7.5:

TABLE 7.5
Total Annual Savings in Chilled Water Consumption
Proposal Ton-days Dollars Acres
Elimination of weekend cooling 1670 $14,529 29.8
Summer night-time air purges 325 $2,828 5.8

Total 1995 $17,357 35.6

These savings represent a reduction of 15.9% in chilled water consumption for annual totals from the
fiscal year 2001.

Furthermore, most of the recommendations presented above involve only changes to code in the
APOGEE control system and thus have no initial capital or personnel costs. Many of the suggestions
are also adaptable to various circumstances that may arise in the future, so that the general principles
may still be applied even if the specific implementations proposed are found to be impractical or
undesirable.

82
 ES 96 Spring 2002

8.0 VENTILATION AND INDOOR AIR QUALITY

83
ES 96 Spring 2002

84
 ES 96 Spring 2002

8.1 Introduction to Indoor Air Quality

In order to provide comfort inside a building – to provide cooling when it is hot outside, and heating
when it is cold – the air supplied into the building must be conditioned. Energy is required to
condition – to cool or to heat – air. Thus, during the majority of time throughout the year, reducing
the rate of outside air supplied into the building would mean reduced energy use to condition the air.
Substantial energy savings are possible by reducing the rate of ventilation in buildings.

However, with reduced ventilation rates, contaminants build up in spaces of a building. For instance,
human respiration produces carbon dioxide, and without adequate ventilation, the resulting high
levels of carbon dioxide can cause discomfort. For other, more toxic compounds, inadequate
ventilation rates could be health, or even life, threatening. Intuitively, people do not paint rooms with
all the windows and doors closed; by opening the doors and windows, the ventilation rate increases,
and provides fresh air.

The class was charged with investigating methods of reducing energy consumption in the buildings
on the Harvard Campus. We knew, intuitively, that by reducing the rate of ventilation in buildings,
we would reduce energy consumption. However, we were also aware that the energy savings could
not come at the expense of the comfort, health, and safety of the occupants in the buildings. We
needed an appropriate balance between saving energy and providing adequate indoor air quality
(IAQ).

Thus, we referred to two different standards in order to determine whether any recommendations we
made would provide adequate indoor air quality. By comparing levels of carbon dioxide (CO2) in the
building with the ventilation rates established and required by the Massachusetts Building Code, we
were able to find a balance between reducing energy use and providing adequate indoor air quality.

85
 ES 96 Spring 2002

8.2 Determining Indoor Air Quality by Carbon Dioxide Levels

8.2.1 C02 in Human Respiration

When humans breathe, air enters the lungs and passes through a series of smaller and smaller tubes
until it arrives at a group of small sacs at the end of the bronchial tubes called alveoli. As blood
passes the alveoli exchange of Oxygen and CO2 occurs by simple diffusion. Oxygen passes from the
alveoli into the blood and CO2 diffuses from the blood into the alveoli. In order for this diffusion
process to occur the concentration of Oxygen must always be higher in the alveoli than in the blood,
and the concentration of CO2 must always be higher in the blood than in the alveoli. This relationship
is maintained by the continual intake of fresh air. The process concludes when the CO2 transferred to
the alveoli is expelled by the lungs into the air. 19

The amount of CO2 in fresh air usually ranges from 350 –450 parts per million. Although CO2 is a
non-toxic gas, levels of 2500 – 5000 ppm can cause headaches. At the extremely high level of
100,000ppm, about 10 percent of people will loose consciousness within ten minutes. Fortunately
levels of 2500-5000 ppm rarely occur within buildings. When the CO2 does reach these levels,
unventilated fuel burning space-heaters (i.e. a kerosene, propane, or natural gas heater) are often the
culprit, although the more serious issue with these space heaters is probably carbon monoxide. It is
also possible to reach high CO2 levels in an unventilated, densely populated, closed space.20

8.2.2 Building Ventilation Requirements

The American Society of Heating Refrigeration and Air-Conditioning Engineers (ASHRAE) bases its
minimum ventilation requirements on the CO2 levels within a building. In “ASHRAE standard 62-
1989, Ventilation For Acceptable Indoor Air Quality”, the standard specifies 15 cubic feet per minute
(cfm) as the minimum amount of outdoor air required per person. The minimum required rate is 30
cfm for a hospital operating room and 60 cfm for a smoking lounge.21

19
Website: Canada Lung Association – Alveoli
20
Website: Greener, Tom. “Indoor Air Quality” May, 1995

86
 ES 96 Spring 2002

Since CO2 is produced during human respiration it is a convenient indicator of the adequacy of indoor
fresh air ventilation rates within occupied buildings. The ASHRAE standard requires HVAC systems
to supply enough fresh air to maintain a building’s CO2 level below 1000 ppm. Thus adequately
ventilated buildings have CO2 levels between the outdoor fresh air C02 level and 1000 ppm. The
average person at rest produces 0.04 kg of carbon dioxide per hour. As occupants breathe throughout
the day a building’s CO2 level will gradually increase, assuming the day begins at low CO2 levels.
Hence most buildings reach their peak CO2 levels late in the afternoon.22

8.2.3 Predictive CO2 Model

We measured the CO2 level at various points throughout Maxwell Dworkin. At many of these points,
we were able to trend the CO2 levels over periods of several days. The measurements suggested that
the average CO2 level for the building was about 500 ppm. Excluding the classrooms on the ground
floor, the CO2 level rarely rose above 600 ppm. From these measurements we concluded that the
building was indeed over ventilated.

Our measurements only provided an accurate indication of the building’s CO2 levels during the
period of the measurements. In order to predict the effects of moderate increases in the building’s
population and quarterly adjustments of the class schedules we needed a tool that would allow us to
predict the CO2 level at any point in the building given a set of parameters. In order to solve this
problem we used the following differential equation to chart the flow of CO2 in and out of a given
space (within a building).

The differential equation used was:

dc(t)/dt =[10 4 N -Q(c(t) - co)]/V (Eq.8.1)

Where:
• c(t) = concentration
† of CO2 within the room (ppm)
• c0 = concentration of CO2 in outside air (ppm)

21
ASHRAE Journal www.ASHRAE.org

87
 ES 96 Spring 2002

• V = volume of the room (ft3)

• Q = air inflow/outflow rate (ft3 / min)
• N = number of people in the room (no units)

The solution to this equation provided the following equation:

CO2 Equation

c(t) = c(O)e-gt + (cO+10 4 N /Q)(1- e-gt) (Eq. 8.2)

Where:
• c(t) = concentration of CO2 within the room (ppm)
†
• c0 = concentration of CO2 in outside air (ppm)

• Q = air inflow/outflow rate (ft3 / min)

• N = number of people in the room (no units)
• g = time constant = Q / V (inverse minutes)• c(0) – represents the initial CO2 level within the
building. Assumed to be 400ppm.

The term c(0) is the initial condition that represents the initial CO2 level within the space under
consideration. It must be supplied when using the equation. For many of our CO2 simulations we
used 450 ppm. This choice was based on measurements within Maxwell Dworkin and should be
adjusted accordingly when applying the equation to a space within another building.

The CO2 equation served as a tool for modeling the CO2 levels within the building. Using this
equation we created a model of the first three floors of Maxwell Dworkin in order to determine the
steady state CO2 levels. We used the current ventilation schedules, and population estimates, and
outside air CO2 levels in order to determine the appropriate values for the equation’s parameters. We
then generated the following model of the CO2 level on Maxwell Dworkin’s first three floors (Figure
8.1).

22
ASHRAE Journal www.ASHRAE.org

88
 ES 96 Spring 2002

CO2 level
(ppm)

Time (minutes)
FIGURE 8.1 CO2 Level Model for Maxwell Dworkin Floors 1,2,3

Clearly the CO2 level reaches a steady state and asymptotes toward 500 ppm within a few hours. This
suggests that the CO2 levels within Maxwell Dworkin will stabilize at around 500 ppm when the
building is operated under current procedures. The model supports the measured average value of
500ppm and suggests that Maxwell Dworkin is indeed over ventilated.

89
 ES 96 Spring 2002

8.2.4 Conclusion and Recommendations

According to the ASHRAE standard an adequately ventilated building must be maintained at a CO2
level below 1000ppm. However it is important to note that ASHRAE selected 1000ppm because it is
the steady state CO2 level when a building is ventilated at 15 cfm per occupant of outside air with a
CO2 concentration of about 350ppm. Thus according to ASHRAE, “indoor carbon dioxide
concentrations in the range of 1000 ppm or less are not related to any health impact from the carbon
dioxide itself, only to odor perception.”23 Hence from a CO2 standpoint building operations managers
should not become alarmed if the CO2 level becomes close to 1000ppm on occasion. In addition in
closed spaces (i.e. classrooms) within a building that maintains the standard 15 cfm ventilation
requirements it should be noted that in the absence of any chemical pollutants the odor, not any health
hazard, represents the primary ventilation concern.

From a CO2 stand point we found Maxwell Dworkin to be highly over ventilated, and recommended a
reduction in the ventilation levels. However odors, and chemical pollutants, are additional factors
that must be considered when determining whether or not a building is over ventilated. In order to
determine the specific reductions for Maxwell Dworkin in terms of cubic feet per minute we had to
examine each of the five air-handling units that contributed to its ventilation system.

8.3 Indoor Air Quality via Massachusetts Building Code

Similar to the ASHRAE Standards 62-1989, the Massachusetts Building Code establishes minimum
ventilation rates. However, the Massachusetts Building Code references a separate ventilation
standard established by the Building Officials Code Administrators (BOCA)24. The ventilation
requirements established by the BOCA International Mechanical Code are binding in the design and
construction of all building in Massachusetts.

The estimated maximum load determines the ventilation rate requirements. Typically, this means
determining the expected occupancy load in the various types of spaces in the building. For instance,
for Office Spaces, the BOCA Code requires a ventilation rate of 20 cubic feet per minute (cfm) per

23
Website: Chiodi, Ron. Confined Space Entry: Electronic Library of Construction, Occupational Safety, and
Health

90
 ES 96 Spring 2002

person. In the design stage of a building, when the actual occupancy is unknown, an expected
maximum occupancy load, as determined by BOCA, is used. If available, more accurate, statistical
information about the actual occupancy may be used. Because we were able to obtain actual
occupancy load data within Maxwell Dworkin, we were able to calculate more accurate safe
ventilation requirements.

The primary concern of the BOCA Code is to provide for the comfort and safety of the occupants; it
is less concerned about energy conservation. Because the class did not want to compromise comfort
and safety levels in order to reduce energy use, we made certain our recommendations satisfied the
ventilation requirements.

Refer to Appendix 8 for a complete list of the ventilation requirements established by the BOCA
International Mechanical Code.

24
Though the Massachusetts Code references Building Officials Code Administrators’ Mechanical Code 1993,
the class relied on the more up-to date International Mechanical Code 2000 for the ventilation requirements.

91
ES 96 Spring 2002

92
 ES 96 Spring 2002

9.0 AIR HANDLING UNITS IN

MAXWELL DWORKIN

93
ES 96 Spring 2002

94
 ES 96 Spring 2002

9.1 Air Handling Units in Maxwell Dworkin

Maxwell Dworkin has 5 Air Handling Units. Combined, they are capable of providing almost 36,500
cfm into Maxwell Dworkin. Individually, the Air Handling Unit serves different areas of the building
(Table 9.1).

TABLE 9.1 Air Handling Units in Maxwell Dworkin

Air Handling Unit Area Served Type of System Capacity (cfm)
AHU 1 Auditorium (G115) SRM 3380
AHU 2 Basement areas SR 2000
AHU 3 Ground Floor SRM 17210
AHU 4 Floors 1, 2, and 3 S 8215
AHU 5 Floors 1, 2, and 3 S 5610
SRM = Supply and Return with Mixed Air System
SR = Supply and Return Air System
S= Supply Air only System

The Air Handling Units in Maxwell Dworkin can be categorized into three types of systems:
• Supply and Return Air System (AHU 2)
• Supply Air Only Systems (AHUs 4 and 5)
• Supply and Return with Mixed Air System (AHUs 1 and 3)

9.2 Method of Determining Ventilation Rates

The Air Handling Units and separate exhaust systems provide for the ventilation in Maxwell
Dworkin. Each of the 5 Air Handling Unit serves a different area of the building. In order to
determine if each of the Air Handling Units was consuming no more energy than necessary while still
providing adequate ventilation, we:

95
 ES 96 Spring 2002

• Calculated the ventilation rate required by the Massachusetts Building Code in the spaces in
served by each Air Handling Unit. For most spaces, this required determining the number of
people expected actually to occupy the different types of spaces in Maxwell Dworkin. Figure
9.1 shows the ventilation requirements of some of the spaces in Maxwell Dworkin. Refer to
Appendix 8 for a complete list.

• Compared our calculated value with the actual ventilation provided by each Air Handling Unit.
For all 5 Air Handling Units, the actual ventilation provided by the Air Handling Unit was at
least slightly greater than the ventilation requirement.

• Measured CO2 levels throughout the building. In order to balance between our goal of reducing
energy consumption and maintaining adequate indoor air quality, we ensured that CO2 levels in
the spaces served by the Air Handling Units remained within the ASHRAE guideline of 1,000
ppm. The CO2 level in occupied spaces throughout Maxwell Dworkin typically ranged from
380 ppm to 600 ppm.

FIGURE 9.1 Excerpt of ventilation requirements, from BOCA International Mechanical Code

96
 ES 96 Spring 2002

In the original design of Maxwell Dworkin, the capacity of the Air Handling Units was determined
primarily on estimated maximum occupant load as determined by the BOCA International
Mechanical Code. Armed with a more accurate estimation of the actual use patterns and occupancy
load in Maxwell Dworkin, we proceeded with our investigation.

9.3 Air Handling Unit 2 – Supply and Return Air System

9.3.1 Description

At a constant flow rate capacity of 2,000 cfm, Air Handling Unit 2 has the smallest capacity of the 5
Air Handling Units in Maxwell Dworkin. It is located in the Room B119 (Mechanical Room) in the
basement, provides ventilation for the basement areas. Appendix 7 has a schematic drawing of Air
Handling Unit 2.

9.3.2 Sequence of Operation

Air Handling Unit 2 consists of a 100% outside air system with an outdoor air damper, pre-filter,
chilled water cooling coil, hot water heating coil and supply fan. The air supplied by Air Handling
Unit 2 is exhausted via one of two ways: 1) the return duct that is an integral part of Air Handling
Unit 2; or 2) separate Exhaust Fans that serve areas with special exhaust requirements. This unit is
controlled through the Siemens APOGEE system’s direct digital control (DDC) interface.

The fan motor in the Air Handling Unit has a Hand-Off-Auto selector switch. Whenever the unit is
commanded to start in either the Hand or Auto position, the outside air damper will open. Only after
the damper is proven open, via a damper end switch, will the fan start. When in the Off position, the
fan shall be off regardless of the command from the DDC system. When in the Auto position, the fan
shall run as scheduled by the DDC system, thus allowing for a time-of-day operation basis.
Furthermore, when in the Auto position, the Air Handling Units operate in Occupied or Unoccupied
mode as follows (Safety mode is also possible, but not relevant as background knowledge to such an
analysis):

97
 ES 96 Spring 2002

Unoccupied:
When indexed to the Unoccupied mode, the supply fan is off. Furthermore, the chilled water coil
valve and outdoor air damper closes (to its minimum position of 20%). The hot water coil valve is
modulated to achieve a minimum air temperature of 50oF to prevent the low temperature thermostat
from tripping.

Occupied:
When indexed to the Occupied mode, the outdoor air damper opens. Once confirmed open, the fan
starts. The heating coil valve and chilled water coil valve modulate in sequence, without overlap, to
maintain the discharge temperature setpoint.

9.3.3 Operation Schedule

Air Handling Unit 2 is set to operate in the Occupied mode 7 days a week, from 7 am to 7 pm. For the
remaining hours, it is set to operate in the Unoccupied mode.

9.4 Method of Investigation

9.4.1 Determination of the Ventilation Rate Requirements

In order to determine the ventilation requirement for the areas served by Air Handling Unit 2, we took
stock of all the types of spaces in the basement. We then determined the ventilation requirement of
each of the types of spaces. For instance, the Teaching Laboratories require 15 cfm/person. With 6
laboratories with an expected maximum occupancy of 9 people per laboratory, we calculated the total
requirement to be 810 cfm. Table 9.2 lists the ventilation requirements of all the spaces located in the
basement.

98
 ES 96 Spring 2002

TABLE 9.2 Ventilation requirement of spaces in the basement of Maxwell Dworkin

Requirment
Type of Space Building Code Requirement Maximum load (cfm)
Mechanical Rooms 0.05 cfm/square feet 5660 feet square 283
Teaching Labs 15 cfm/person 9 people x 6 laboratories 810
Offices 20 cfm/person 2 people x 4 offices 160
Bathrooms 75 cfm/urinal or water closet 8 urinals 600
Electrical Closet 0.05 cfm/square feet 220 square feet 110
TOTAL 1963

9.4.2 Comparison with the Actual Ventilation Provided

As previously mentioned, Air Handling Unit 2 provides a constant supply flow rate of 2,000 cfm in
the occupied mode. The required supply ventilation rate of 1,963 cfm closely matched the capacity of
Air Handling Unit 2.

9.4.3 Measurement of CO2 Levels

We took measurements of CO2 levels throughout the spaces typically occupied by people in the
basement. The CO2 levels ranged between 380 ppm and 600 ppm, well below the 1,000 ppm
recommended by the ASHRAE Standards 62-1989.

In the Mechanical Rooms, the CO2 levels were higher. The Code permits drastically lower
ventilation rates in the Mechanical Rooms because people do not occupy these spaces for extended
periods of time. Thus, while the Mechanical Rooms account for about 33% of the floor area in the
basement – approximately 5,660 square feet out of a total of 16,622 square feet – they require only
283 cfm. Because of the lower ventilation rate, these spaces obviously had higher CO2 levels than the
occupied spaces. In comparison, the two bathrooms require 600 cfm by code.

99
 ES 96 Spring 2002

9.5 Findings

Based purely on measurements of CO2 levels in the occupied spaces in the basement, we were
tempted to recommend reducing the flow rate of Air Handling Unit 2 in order to reduce energy
consumption in Maxwell Dworkin. However, for two reasons, we determined that reducing the flow
rate of Air Handling Unit 2 would be inadvisable.

First, we refused to make any recommendations that did not comply with the Massachusetts Building
Code. Secondly, because of vast disparity in the ventilation requirements of substantial areas of the
spaces in the basement, we determined that readings of average CO2 levels could not be relied upon
to ensure that adequate ventilation was provided to all of the occupied spaces.

Thus, we found that Air Handling Unit 2 was operating as efficiently as permitted by the
Massachusetts Building Code.

9.6 Exhaust Fans

9.6.1 Overview

Certain spaces in the building have special exhaust requirements. The Massachusetts Building Code
requires that the exhaust air from such spaces be vented directly to the outside25.

A total of 9 Exhaust Fans service Maxwell Dworkin. Of the 9 fans, Exhaust Fans 1, 2, 3, and 4 serve
the following areas throughout all of Maxwell Dworkin:
• Electrical Closets
• Photocopy Rooms
• Bathrooms
• Janitor’s Closets
• Kitchenettes

25
780 CMR: State Board of Building Regulations and Standards: The Massachusetts State Building Code

100
 ES 96 Spring 2002

These 4 Exhaust Fans have the most significant impact on the day-to-day operation of the Air
Handling Units.

Exhaust Fans 5 and 6 are for the two fume hoods, and only one of the two fans is currently operating.
Exhaust Fans 7, 8, and 9 operate only to remove smoke in the stairwells in case of a fire emergency.
Because Exhaust Fans 5 through 9 have minimal effects on the daily operation of the ventilation
system, our investigation focused exclusively on Exhaust Fans 1, 2, 3, and 4.

9.6.2 Operation of the Exhaust Fan

The only settings for the Exhaust Fans are ON and OFF, and the fans are controlled by Siemens
APOGEE system’s direct digital control (DDC) interface. They operate 7 days a week, 24 hours a
day.

9.7 Method of Investigation

9.7.1 Exhaust Provided vs. Exhaust Requirements

We identified all the spaces throughout Maxwell Dworkin that were served by Exhaust Fans 1, 2, 3,
and 4, and summed up the exhaust requirements. We disaggregated the exhaust requirements not
only by individual Exhaust Fans, but also by the areas each Exhaust Fan served (Table 9.3).

101
 ES 96 Spring 2002

Capacity 1st 2nd 3rd Potential

EF Areas served (cfm) Basement Ground FL FL FL Total Reduction

Electrical Closets /
1 Copy rooms 1450 0 30 170 110 110 420 1030
Toilets / Janitor's
2 Closets 5480 600 825 675 675 675 3450 2030

3 Kitchenettes 400 0 25 25 25 25 100 300

4 Elect. Closets 1250 110 25 25 25 25 210 1040

Total 8580 710 905 895 835 835 4180 4400

TABLE 9.3 Exhaust requirements (cfm) for spaces with special exhaust requirements in Maxwell Dworkin

We then determined the actual exhaust provided by the 4 Exhaust Fans, and compared the two values.

9.7.2 Findings

We found that each of the 4 Exhaust Fans were exhausting more air than the requirement. Exhaust
Fans 1 and 4 provided for more than 1,000 cfm of exhaust above and beyond the requirements of the
spaces they served, while Exhaust Fan 2 provided for more than 2,000 cfm of excess exhaust.

Specifically in the basement of Maxwell Dworkin, the Exhaust Fans exhausts air from the bathrooms
(600 cfm) and the electrical closets (110 cfm), providing a total of 710 cfm of exhaust. Thus, of the
2,000 cfm supplied by Air Handling Unit 2, the return duct and fan accounted for 1,290 cfm of
exhausted air, while the Exhaust Fans accounted for about 710 cfm.

Figure 9.2 shows the flow rate of supply air provided by the Air Handling Units (the blue boxes), the
flow rate of air exhausted by various means (the darker green boxes), and the flow rate of exhaust air
required in the spaces with special exhaust requirements in Maxwell Dworkin (the lighter green
boxes). The 4 Exhaust Fans are represented by a single darker green box on the top of the diagram.
Throughout all of Maxwell Dworkin, the 4 Exhaust Fans provide 8,580 cfm of exhaust for all of the
spaces with special exhaust requirements. However, a total of only 4,180 cfm of exhaust is required
in these spaces. Therefore, the rate of exhaust could potentially be reduced by 4,400 cfm.

102
 ES 96 Spring 2002

For Air Handling Unit 2, the Exhaust Fans provide added capacity to remove the exhaust air from the
basement (since Air Handling Unit 2 has the capacity to remove all 2,000 cfm of the supply air it
provides). However, in the operation of Air Handling Units 4 and 5 – the supply air only systems
which have no capacity for the exhaust of air – the Exhaust Fans play a much more significant role.

Although we found that there was the potential to substantially reduce the flow rate of Exhaust Fans
1, 2, 3, and 4, such reductions in and of themselves would have resulted in negligible energy and
monetary savings. However, these potential reductions in the flow rate of the Exhaust Fans in
conjunction with our findings about Air Handling Units 4 and 5 showed that substantial reductions in
energy consumption could be achieved without compromising the comfort or air quality in Maxwell
Dworkin.

103
 ES 96 Spring 2002

FIGURE 9.2 Supply and exhaust of air in Maxwell Dworkin

104
 ES 96 Spring 2002

9.8 Air Handling Units 4 and 5 – Supply Air Only Systems

Air Handling Units 4 and 5 are located on the roof of Maxwell Dworkin and are responsible for
providing the ventilation to floors 1, 2 and 3. Air Handling Unit 4, the larger of the two, has a current
flow rate capacity of 8,215 cfm. Its supply fan is serviced by a 10 hp motor. Air Handling Unit 5 has
a current flow rate capacity of 5,610 cfm, with a supply fan serviced by a 7.5 hp motor. Please refer to
Appendix 7 for schematic drawings of Air Handling Units 4 and 5.

9.8.1 Sequence of Operation

Though Air Handling Units 4 and 5 differ in capacity, their sequences of operations’ are
indistinguishable. Both Air Handling Units consist of a 100% outside air system with an outdoor air
damper, pre-filter, chilled water cooling coil, hot water heating coil and supply fan – no capacity for
return air is provided by the Air Handling Units. The units are controlled through the Siemens
APOGEE system’s direct digital control (DDC) interface.

The fan motor in Air Handling Units 4 and 5 has a Hand-Off-Auto selector switch. Whenever either
of the units is commanded to start in either the Hand or Auto position, the outside air damper opens.
Only after the damper is proven open, via a damper end switch, will the fan start. When in the Off
position, the fan shall be off regardless of the command from the DDC system. When in the Auto
position, the fan shall run as scheduled by the DDC system, thus allowing for a time-of-day operation
basis. Furthermore, when in the Auto position, the Air Handling Units operate in Occupied or
Unoccupied mode as follows (Safety mode is also possible, but not relevant as background
knowledge to such an analysis):

Unoccupied:
When indexed to the Unoccupied mode, the supply fan is off. Furthermore, the chilled water coil
valve and outdoor air damper closes (to its minimum position of 20%). The hot water coil valve is
modulated to achieve a minimum air temperature of 50oF to prevent the low temperature thermostat
from tripping.

105
 ES 96 Spring 2002

Occupied:
When indexed to the Occupied mode, the outdoor air damper opens. Once confirmed open, the fan
starts. The heating coil valve and chilled water coil valve modulate in sequence, without overlap, to
maintain the discharge temperature setpoint – currently set to 68oF regardless of season.

9.8.2 ‘Original’26 Operation Schedule

As per the sequences of operations, the ‘original’ operation schedules of Air Handling Units 4 and 5
are identical. Air Handling Units 4 and 5 were set to operate in the Occupied mode 7 days a week,
from 7 am to 7 pm. For the remaining hours, Air Handling Units 4 and 5 were set to operate in the
Unoccupied mode.

9.9 Method of Investigation

9.9.1 Verification of Operation Sequence

The operation sequences described above were verified to be in accordance with experimentally
obtained measurements. Using Siemens’ APOGEE interface, the discharge air temperature of Air
Handling Units 4 and 5 were trended over a two week period. The Figures 9.3 and 9.4 represent a
typical trend for a day in the two week period:

26
The group of settings that were in effect at the beginning of our investigation into Maxwell Dworkin – i.e.
that group of settings that were in effect in mid-February 2002.

106
 ES 96 Spring 2002

Discharge Air Temperature - AHU 4 Discharge Air Temperature - AHU 5

69 70

68 68

Temperature (deg. F)
Temperature (deg. F)

67 66

66 64

65 62

64 60

63 58

12:00:00

14:24:00

16:48:00

19:12:00

21:36:00
0:00:00

2:24:00

4:48:00

7:12:00

9:36:00

0:00:00
0:00:00

2:24:00

4:48:00

7:12:00

9:36:00

0:00:00
12:00:00

14:24:00

16:48:00

19:12:00

Time (3/13/02) 21:36:00 Time (3/13/02)

FIGURE 9.3 Trend for AHU 4 FIGURE 9.4 Trend for AHU 5

As is apparent from the above figures, the 7am to 7pm Occupied mode operation appears to be in
effect. Furthermore, the discharge air temperature remains within the range of 68oF ± 1oF – a constant
discharge air temperature of 68oF may, thus, be assumed.

Another crucial value in our analysis of Air Handling Units 4 and 5 was the flow rates at which they
operated. In order to confirm that these units were providing supply air at full capacity, we measured
the actual flow rate of air inside the supply duct of Air Handling Unit 4, the larger of the two units.
By “traversing the duct”27 of the supply air duct, we measured an average flow rate of 9,373 cfm.
Our measurement was actually 14% higher than the rated capacity of 8,215 cfm of Air Handling Unit
4. We attributed this disparity to the fact due to physical constraints, we could only take
measurements in a location in the duct expected to have high turbulence. Regardless, the fact that
our measurement was higher than the rated flow rate of Air Handling Unit 4 indicated that it was
operating at full capacity. Please refer to Appendix 5 for measurements and calculations.

Although we attempted to repeat the measurement for Air Handling Unit 5, the only reasonable
section in which measurements could be taken was enveloped in a wall, and thus inaccessible.

27
A standard method of determining the flow rate of air in a duct by taking a series of measurements of the flow
rate inside the duct at strategic locations, and then taking the average of all the readings.

107
 ES 96 Spring 2002

9.9.2 Cost of Operation

In an attempt to ascertain the annual cost of operating Air Handling Units 4 and 5, it became
necessary to calculate the approximate annual cost of conditioning and circulating 1 cfm of fresh air.
Treating Air Handling Unit 4 and 5 as one unit with a 13,825 cfm flow rate capacity, the following
assumptions were made during calculation:

ß Air Handling Units 4 and 5 are on a 7 days a week, 7am – 7pm schedule

ß The supply fan motor was rated at 17.5 hp.

ß All the power was delivered to, and used by the supply fan.

ß No compensation was made for humidity levels during the summer.

ß From mid-June to mid-October, no hot water is supplied to Maxwell Dworkin.

ß Energy transfer from the heating/cooling coils to the incoming air in the Air Handling Unit
was 100% efficient and obeyed the following relationship:

U = mcDT (Eq. 9.1)

where:

U is the energy (BTU) required to change the temperature by DT (oF) of air with
weight m (lbs) and specific heat capacity c (BTU / lboF).

Through use of a weather data set for the year 2001, the cost of conditioning and circulating one cfm
per year was calculated, under the above assumptions, to be $2.55. Please see Appendix 5 for the
complete calculation. Using this calculated annual cost per cfm, the operating costs of Air Handling
Units 4 and 5 can be calculated to ~$32,300.

For reasons of comparative accuracy, it may be noted that $4.30 is a figure used by Harvard
Maintenance for this very purpose – however, this figure assumes Occupied mode operation 7 days a
week, 24 hours a day. The difference in figures (if $4.30 is converted to the same schedule of
operation that is) may, in part, be attributed to the circulation assumptions made above. In practice, a
fan’s motor is typically overrated by a factor of 20-60% - thus, the assumption of the fan running at
17.5 hp may be thought of as an upper bound on the circulating costs.

108
 ES 96 Spring 2002

9.9.3 Analysis of Space Serviced

When a building is designed, systems such as ventilation are determined on the basis of an expected
maximum occupancy load. However, as shall be seen to be the case in Maxwell Dworkin, the actual
occupancy load of the building is often quite different from the ‘designed for’ occupancy load.

Air Handling Units 4 and 5 must supply the amount required by the occupancy load on floors 1, 2 and
3 as well as the amount required by the Exhaust Fans servicing special areas on floors 1, 2 and 3.

From ‘original’ operation values:

Supply rate of Air Handling Units 4 and 5 = 13,825 cfm

Discharge rate of Exhaust Fans on floors 1, 2 and 3 = 7,120 cfm

Therefore, supply rate for occupancy load = 6,705 cfm

In determining the number of people that such a supply rate may be deemed to supply adequate
ventilation for, the ventilation requirements cited by Massachusetts Building Code must be referred
to. Noting that the Massachusetts Building Code references the Building Official Code
Administrators (BOCA) International Mechanical Code for determining ventilation requirements, a
supply rate of 20 cfm of fresh air per person in an office space environment can be established.

Using 20 cfm per person (note that Air Handling Units 4 and 5 are 100% outside air systems), a
supply rate of 6,705 cfm can be calculated to represent adequate ventilation for 335 people.

9.10 Findings

9.10.1 Actual Occupancy Load

In Maxwell Dworkin, a current list of all faculty, staff and graduate students assigned to space on
floors 1, 2 and 3 is kept by the DEAS Facilities Office. For the academic year 2001-2002, 120 people
are assigned to space on these floors.

109
 ES 96 Spring 2002

For proceeding calculations a 50% safety margin was added on top of the 120 person documented
occupancy, making the actual occupancy 180 people. This safety margin was intended to take into
account any occupancy flux due to traffic specifically from the seminar rooms or common area
lounges.

At 20 cfm per person, the supply rate required for the actual occupancy can be calculated to be equal
to 3600 cfm.

9.10.2 Total Requirements

From previously calculated values:

Discharge rate required by Exhaust Fans on floors 1, 2 and 3 = 2,565 cfm

Supply rate required by occupancy load on floors 1, 2 and 3 = 3,600 cfm

Therefore, supply rate required by Air Handling Units 4 and 5 = 6,165 cfm

This calculated required supply rate for Air Handling Units 4 and 5 corresponds to 55.4 % reduction
from the ‘original’ supply rate. Furthermore, at the expected occupancy load of 120 people on floors
1, 2 and 3 in Maxwell Dworkin, the calculated rate of supply for Air Handling Units 4 and 5 provides
30 cfm per person – a 50 % higher rate of ventilation than that recommended by the Massachusetts
Building Code.

9.10.3 Impact of Reductions

In considering the possibility of reducing the supply rate of Air Handling Units 4 and 5, the impact on
the CO2 level on floors 1, 2 and 3 was predicted using the previously established CO2 differential
model. Given an occupancy load of 180 people on floors 1, 2 and 3 as well as outside CO2
concentration levels of 400 ppm, Figure 9.5 below represents the expected CO2 behavior:

110
 ES 96 Spring 2002

FIGURE 9.5 CO2 model with a 180 person occupancy on Floors 1, 2 and 3.
Outside CO2 level as well as initial CO2 level was assumed to be 400 ppm

The CO2 model suggests that even in the worse case scenario of all 180 people remaining on floors 1,
2 and 3 from 7 am to 7 pm, the CO2 level will only have reached 680 ppm by 7 pm – a perfectly
adequate level of CO2.

Another factor that must be taken into account is the energy content of the air supplied by Air
Handling Units 4 and 5. Noting that the building setpoint temperature of Maxwell Dworkin is 72oF
regardless of season, and that air is supplied by Air Handling Units 4 and 5 at a temperature lower
than this, it becomes evident that the supply air possesses an amount of ‘free’ cooling energy.
However, by decreasing the quantity of total supply air, this amount of ‘free’ cooling energy is also,
proportionally, decreased. In order to compensate for this energy offset, we recommend season
specific solutions.

9.10.4 Winter Month Operation

During the winter months – approximated to be mid-October to the beginning of May – the
temperature of the discharge air from Air Handling Units 4 and 5 may be lowered in a manner that
compensates for the aforementioned energy offset. In solving this problem, the following equations
can be written (assuming original discharge temperature to be 67oF), with U representing total energy
in the supply air:

111
 ES 96 Spring 2002

U original = U reduced
fi moriginal DToriginal = mreduced DTreduced
13,825 o
fi DTreduced = •4 F
6,165
fi DTreduced = 8.97 o F

Therefore, by lowering the discharge temperature of Air Handling Units 4 and 5 to approximately
63oF, the energy offset will be compensated for during the winter months. Furthermore, due to the
now decreased temperature differential between the outside air and discharge air of Air Handling
Units 4 and 5, the annual cost of conditioning a cfm is reduced. In fact, if the calculations in
Appendix 5 are run with a discharge temperature of 63oF during the months mid-October to the
beginning of May, a new total annual cost per cfm of $2.32.

9.10.5 Summer Month Operation

Though lowering the discharge air temperature for Air Handling Units 4 and 5 may be an
economically beneficial solution during winter months, a similar decrease during the summer months
may not be so advantageous. It is, thus, with this idea in mind that we pose an alternative solution.
Reflecting back to the psychrometric chart (Figure 9.6), it may be observed that the summer building
setpoint temperature may be raised to correspond with the summer comfort zone.

112
 ES 96 Spring 2002

FIGURE 9.6 Psychrometric chart adopted from ASHRAE Fundamentals 2001

In Figure 9.6, the blue dot represents the location of Maxwell Dworkin’s average winter day’s
environment. Similarly, the red dot represents the location of Maxwell Dworkin’s average summer
day’s environment. If the red dot is considered alongside the summer comfort zone (the hashed region
represent by ‘Summer’), it becomes evident that Maxwell Dworkin currently operates at an
uncomfortable summer environment. Thus, in an attempt to negate the aforementioned energy offset,
the summer-time building setpoint temperature may be increased to 76oF – a temperature that, on an
average summer day in Maxwell Dworkin, will place the environment near the middle of the summer
comfort zone.

113
 ES 96 Spring 2002

9.11 Recommendations

9.11.1 Reduced Ventilation

From the previous calculations, it may be concluded that Maxwell Dworkin is supplying ventilation
for a much larger occupancy load than currently exists.

In summary:

Exhaust Fans (all floors) are only required to discharge: 4,180 cfm
Exhaust Fans (floors 1, 2 and 3) are only required to discharge: 2,565 cfm
Exhaust Fans (all floors) originally discharged: 8,580 cfm

Air Handling Units 4 and 5 are only required to supply: 6,165 cfm
Air Handling Units 4 and 5 originally supplied: 13,825 cfm

Furthermore, Air Handling Units 4 and 5 operate on the same schedule throughout the week. This
schedule of operation neglects ventilation reductions that may be obtained from the considerably
reduced weekend occupancy load.

Given that, for most, work on weekends is optional, mornings and late afternoons are less likely to be
busy. Furthermore, much lower total occupancy loads are expected over the weekend. One solution to
this problem that we recommend is to reduce ventilation over the weekends by taking advantage of
the scheduling functionality of Siemens’ APOGEE system. A possible weekend schedule, shown in
Table 9.4, for Air Handling Units 4 and 5 would see the units operating in Occupied mode for a total
of 8 hours over the previous 24 weekend Occupied hours.

114
 ES 96 Spring 2002

Mode Air Handling Unit 4 Air Handling Unit 5

Occupied 7am – 8am 7am – 8am
(Maxwell Dworkin placed at (Maxwell Dworkin placed at
setpoint) setpoint)
11am – 1pm 11am – 1pm
(Peak occupancy load (Peak occupancy load
ventilation) ventilation)
4pm – 5pm 4pm – 5pm
(Final ventilation for the day) (Final ventilation for the day)

Unoccupied Remaining time Remaining time

TABLE 9.4 Possible weekend schedule for AHUs 4 and 5

Though the functionality of this schedule has not been experimentally verified, observation of
weekend occupancy loads suggests that it will be supply more than adequate ventilation. Furthermore,
if this schedule is combined with summer nighttime air purges, Air Handling Units 4 and 5 may
commence their weekend ventilation at a later hour in the morning, thus taking advantage of the ‘free’
cool air brought in during the night.

9.11.2 Implementation

In order to decrease the supply rate of Air Handling Units 4 and 5 as well as the discharge rate of the
Exhaust Fans servicing floors 1, 2 and 3, the fan speed of the various units will have to be adjusted.
To this problem, we propose three solutions.

Variable Frequency Drives:

By installing variable frequency drives on an existing fan motor, smooth control of that unit’s fan
speed can be achieved. Given that the Exhaust Fans service areas of defined, constant requirements,
such control would not be appropriate in their case. However, we feel that Air Handling Units 4 and 5

115
 ES 96 Spring 2002

may benefit from such a modification. The variable frequency drives would allow the air supply of
Air Handling Units 4 and 5 to be fine tuned, thus optimizing required ventilation for floors 1, 2 and 3.

Cost estimates obtained from Mr. Jay Phillips – Manager of the FAS Building Systems – show that a
complete retrofit of a 10 hp existing motor with a variable frequency drive will lie in the region of
$14,000 to $15,000.

Pulley Ratio change:

The second, simpler suggestion relies on the fact that all of the fans are run off of a pulley belt –
observation confirms this fact. By changing the pulley ratio of the fan’s pulley mechanism, the fan
speed can be lowered to the desired speed. Given that the Exhaust Fans, as previously mentioned,
service areas of defined, constant requirements, such a change would be very appropriate.
Furthermore, this method is also directly applicable to Air Handling Units 4 and 5 given the fact that
the occupancy load will never be expected to rise above 180 people.

Cost estimates show that labor and parts for such a modification lie in the region of $1,000.

Alternating Schedule Operation:

Our last suggestion, initially implemented as a test method, again takes advantage of the scheduling
functionality of Siemens’ APOGEE system. By balancing the time that either Air Handling Unit 4 or
5 operates in the Occupied mode, a predetermined level of average ventilation may be obtained. The
schedule shown in Table 9.5 was put into operation in Maxwell Dworkin and observed over a 2.5
week period:

Mode Air Handling Unit 4 Air Handling Unit 5

Occupied 7am – 10am 7am – 8am
1:45pm – 5pm 9:45am – 2pm
4:45pm – 7pm

Unoccupied Remaining time Remaining time

TABLE 9.5 Alternating operation schedule of AHUs 4 and 5

116
 ES 96 Spring 2002

During the period of observation, CO2 levels were measured to stay within acceptable levels – in fact,
CO2 levels showed very little change from previously recorded values: common areas were typically
measured between 400 ppm and 600 ppm; passageways ranged from 380 ppm to 550 ppm, and
offices ranged from 450 ppm to 700 ppm.

Given that the Exhaust Fans servicing floors 1, 2 and 3 were not modified, the ventilation supplied
during this test period was significantly lower than that recommended by our previous calculations.
However, no complaints were voiced by the occupants of floors 1, 2 and 3 or by the Air Handling
Unit managers. Nevertheless, if this operation schedule is to be implemented for long run use, we
recommend that the discharge rate of the Exhaust Fans be changed to the previously calculated
value.

9.11.3 Savings

In considering total savings obtainable from the recommendations hitherto, three main contributors
may be taken into account:

• Reducing the constant ventilation rate of Air Handling Units 4 and 5 corresponds to a reduction
in annual running costs.
• Reducing the total ventilation time on weekends corresponds to a similar reduction.
• Reducing the discharge air temperature of Air Handling Units 4 and 5 during winter months
corresponds to a reduction in cost of conditioning per cfm per year.

117
 ES 96 Spring 2002

In economic terms:

Recommendation Savings ($)

Set combined ventilation rate of Air Handling Units 4 and 5 to 6,165 cfm. 19,533.0028
Adopt suggested weekend schedule. 2,994.0029
Decrease discharge air temperature to 63oF during winter months. 1,418.0030

TOTAL 23,945.00

TABLE 9.6 Operation recommendations for AHUs 4 and 5, and associated savings

9.11.4 Payback Period

Simple payback periods for the implementations suggested were calculated and are presented below:

Implementation Method Implementation Cost Simple Payback

31
($) Period (Years)
Variable frequency drive installation on Air Handling 90,000.00 3.8
Units 4 and 5 and Exhaust Fans.
Variable frequency drive installation on Air Handling 34,000.00 1.4
Units 4 and 5 and pulley ratio change on Exhaust Fans.
Pulley ration change on Air Handling Units 4 and 5 and 6,000 0.3
Exhaust Fans.
Alternating schedule operation of Air Handling Units 4 and 4,000 0.2
5 and pulley ratio change on Exhaust Fans.

TABLE 9.7 Implementation costs and payback period of operation recommendations of AHUs 4&5

28
Based on a reduction of 7,660 cfm at $2.55 annual cost per cfm
29
There are 68 total Occupied mode week hours under this schedule.
30
Calculations show an approximate $0.23 annual saving per cfm under such a discharge temperature reduction.
31
Based on annual savings of $23,760.00

118
 ES 96 Spring 2002

9.12 Air Handling Units 1 and 3 – Supply and Return with Mixed Air
Systems

The third type of air handling units servicing Maxwell Dworkin are those with mixed air functionality
which service the entire ground floor. This functionality is an energy saving device that takes return
air from the ground floor, and then depending on the temperature and humidity of the room and of
outside air, determines most efficient combination of return and outside air and sets the damper
positions accordingly to create the desired mix of air. The mixed air is then conditioned, either heated
or cooled to maintain a supply temperature, and passed back into the area by a supply fan.

9.12.1 Sequence of Operation

Unlike the supply air only and the supply and return air systems, the sequence of operations for Air
Handling Units 1 and 3 are quite different. Similarities can be seen, however, in the basic path each
follows from unoccupied to occupied mode.

Unoccupied:
When indexed to the Unoccupied mode, the chilled water valve is turned off and heating is only used
to ensure that the space temperature in the Air Handling Unit does not drop below 50 degrees to
prevent the low temperature thermostat from tripping and the chilled water coils from breaking. The
fans will only run in those cases when the hot water valve is opened to prevent such an event.

Cool Down and Warm Up Modes

When passing from the Unoccupied mode to the Occupied mode, the air handling unit may be
indexed to a Cool Down or Warm Up mode. If the return air temperature is 2 degrees greater than the
return air setpoint, the air handling unit will enter into Cool Down mode where it will modulate the
supply air setpoint to achieve a supply air temperature of 55 degrees. However, if the return
temperature is 2 degrees less than the return air setpoint, the system will enter Warm Up mode and
close the outside air dampers to 20% and reheat the return air. Both of these modes are enacted to
quickly restore the return temperature to the desired return air setpoint of 72 degrees. Once that
temperature has been reached, the system will settle into Occupied mode. If neither cool down or
warm up mode was necessary, the air handling unit will immediately go into Occupied mode at from
Unoccupied mode.

119
 ES 96 Spring 2002

Occupied Mode
In Occupied Mode, both air handling units behave quite differently. Of particular note is the presence
of a carbon dioxide monitor in Air Handling Unit 1 which is absent in Air Handling Unit 3.
Therefore, the Occupied mode for Air Handling Unit 1 checks not only the temperature and humidity
of the room and outside air, but also carbon dioxide levels of the return air to determine the damper
positions. Air Handling Unit 3 is unlike any other in that it possesses a variable frequency drive and
variable air volume rate functionality. Furthermore, it attempts to maintain a discharge air
temperature of 65 degrees whereas the four other air handling units supply air at 68 degrees. Further
description of the Occupied modes of these air handling units will be included with the study of each
particular air handling unit.

9.12.2 Original32 Operation Schedule

As per the sequences of operations, the ‘original’ operation schedules of Air Handling Units 1 and 3
are identical. Air Handling Units 1 and 3 were set to operate in the Occupied Mode 5 days a week,
Monday through Friday, from 7 am to 7 pm. For the remaining hours, Air Handling Units 1 and 3
were set to operate in the Unoccupied mode.

32
The group of settings that were in effect at the beginning of our investigation into Maxwell Dworkin – i.e.
that group of settings that were in effect in mid-February 2002

120
 ES 96 Spring 2002

9.13 Method of Investigation

9.13.1 Verification of Operation Sequence

The operation sequences described above were verified to be in accordance with experimentally
obtained measurements. Using Siemens’ Apogee interface, the discharge air temperature of Air
Handling Units 1 and 3 were trended over a two week period. The Figures 9.7 and 9.8 represent a
typical trend for a day in the two week period:

Figure 9.7 Trend for AHU1 Figure 9.8 Trend for AHU 3

The above figures show that discharge air temperature for Air Handling Unit 1 remains within the
range of 70oF ± 1.5oF and for Air Handling Unit 3, the discharge air temperature is 65oF ± 1oF.
Therefore, a constant discharge air temperature of 70oF for Air Handling Unit 1 and 65oF for Air
Handling Unit 3 may be assumed.

121
 ES 96 Spring 2002

9.13.2 Cost of Operation

To obtain the annual cost of operating Air Handling Units 1 and 3, we again calculated the
approximate annual cost of conditioning and circulating 1 cfm of mixed air. Beginning with Air
Handling Unit 1, the following assumptions were made for calculation:
• Air Handling Unit 1 is on 5 days a week from 7 am to 7 pm
• All the power was delivered to and used by the supply fan
• No compensation was made for humidity during the summer months
• From mid-June to mid-October, no hot water is supplied to Maxwell Dworkin
• Energy transfer from the heating/cooling coils to the incoming air in the air handling units was
100% efficient and obeyed the relationship described in Eq. 9.1:
U = mcDT
In addition, for Air Handling Unit 1, the assumption was made that the supply fan motor was rated at
5 horsepower and the volume flow rate of the system was 3,380 cfm.
For Air Handling Unit 3, the supply fan motor has a maximum rating of 25 horsepower and a
maximum volume flow rate of 17,200 cfm. However, several VAV boxes have been uninstalled or
turned off, so that the current maximum volume flow rate is 16,512 cfm. For the purposes of our
calculations, we will use trended data which shows the system providing an average of 9,000 cfm to
the ground floor. A corresponding reduction in energy to the supply fan could result in an estimated
energy consumption of about 12 horsepower.

122
 ES 96 Spring 2002

Trended data of mixed air temperature was also used in computing the cost of conditioning and
circulating one cfm of mixed air per year under the above assumptions. Figures 9.7.1 and 9.8.1
represent a typical trend for a day in the two-week period during April:

FIGURE 9.7.1 Trend for AHU 1 FIGURE 9.8.1 Trend for AHU 3

The above figures show that mixed air temperature for Air Handling Unit 1 remains within the range
of 62oF ± 2oF and for Air Handling Unit 3, the mixed air temperature is 62 oF ± 1.5 oF. Therefore, a
constant mixed air temperature of 62 oF for both air handling units may be assumed.

Total cost of conditioning and circulating one cfm was determined from similar data to be $1.86 for
Air Handling Unit 1 and $1.82 for Air Handling Unit 3. For complete calculations, refer to Appendix
6. Using these calculated annual costs per cfm, the operating cost of running Air Handling Unit 1 is
determined to be ~$6,300 and the operating cost of Air Handling Unit 3 to be ~$16,380. However, as
explained in section 9.9.2, a fan’s motor is typically overrated by a factor of 20-60%, so the
assumption of the fan in Air Handling Unit 1 running at 5 hp or the fan in Air Handling Unit 3
running at a maximum of 25 hp may be thought of as an upper bound on circulating costs.

123
 ES 96 Spring 2002

9.13.3 Analysis of Space Served

Air Handling Units 1 and 3 are faced with the same occupancy load issues as Air Handling Units 4
and 5. The original designed occupancy load for the ground floor is quite different than actual
occupancy load.

Air Handling Unit 1 must provide the volume flow rate required by the occupancy load in Maxwell
Dworkin G115.

From ‘original’ operation values:

Supply rate of Air Handling Unit 1 for occupancy load = 3,380 cfm
Referencing once again the BOCA International Mechanical Code a supply ventilation rate of 15 cfm
per person may be established for an auditorium space such as the seminar room G115.

Using 15 cfm per person, a supply volume flow rate of 3,380 cfm may be determined to provide
adequate ventilation for 225 persons.

Air Handling Unit 3 provides ventilation to the remainder of the ground floor. Since it is a variable
volume flow rate system, we will maintain our original trended data average of 9,000 cfm.

From ‘original’ operation values

Supply rate of Air Handling Unit 3 for occupancy load = 9,000 cfm
The ground floor is mainly comprised of classrooms, corridors, and BOCA recommends 15 cfm per
person

Assuming an average of 15 cfm per person is adequate for all areas of the ground floor, a supply
volume flow rate of 9,000 cfm may be determined to provide adequate ventilation for 600 persons.

124
 ES 96 Spring 2002

9.14 Findings for Air Handling Unit 1

9.14.1 Understanding Air Handling Unit 1

Air Handling Unit 1 is unique in that it possesses a carbon dioxide monitor located in the return air
duct. When this air handling unit is indexed to the Occupied mode, carbon dioxide levels are
checked before any other inputs. If the maximum carbon dioxide level has been detected, the air sent
to the room will come completely from outdoor air, which is likely more costly to condition. During
our studies, the air handling unit was set to bring in full outside air when carbon dioxide levels
reached 750 ppm and then return to normal operating conditions once they had dropped below 700
again. However, the American Society of Heating Refrigeration and Air Conditioning Engineers
classifies carbon dioxide readings of 1000 ppm to be comfortable, so the current maximum of 750
ppm seems rather low. We suggest raising the maximum carbon dioxide level to 1000 ppm or even
higher in order to minimize the number of times that Air Handling Unit 1 will be sent into the
inefficient mode of accepting 100% outside air. Once carbon dioxide levels are measured, indoor and
outdoor enthalpy is measured and if the outdoor enthalpy is much higher than indoor enthalpy,
dampers will be closed to their minimum position to return as much indoor air as possible. If outdoor
enthalpy level is acceptable, the system will settle into the economizer program. The economizer
program is a simple code which looks at indoor and outdoor temperatures and determines the most
energy efficient damper positions and sets them accordingly. The current setting of the operating
sequence states that when the air handling unit goes in to Occupied mode, it will stay there until 7 pm,
at which point the system will return to Unoccupied mode.

9.14.2 Actual Occupancy Load

In Maxwell Dworkin, a current listing of classes assigned to room G115 is kept in the DEAS
Facilities Office. Class sizes for an average semester range in size from 5 to about 120 students, of
course this is assuming all students attend lectures. For proceeding calculations, a safety margin of
25% over the 120 students observed in the classroom will be used. This safety margin (150 students)
was intended to account for semesters that may have unusually large classes. At 15 cfm for an
auditorium space, the actual ventilation requirement for the occupancy is calculated to be 2,250 cfm –
well below the 3,380 cfm currently supplied. Occupancy numbers and ventilation rates were verified

125
 ES 96 Spring 2002

to be in accordance with experimentally obtained measurements by measuring carbon dioxide levels.

We observed carbon dioxide readings in the room to only peak above this level once or twice during
the day and quickly drop to a much lower level of 400 ppm.

FIGURE 9.9 Carbon Dioxide Trend for AHU 1

Figure 9.9 shows an example of how the CO2 levels during a typical class day in room G-115 reach
800 ppm only once, at which point full outside air is brought into the room and carbon dioxide levels
quickly drop to back to 400 ppm. The small bump in the graph that appears between 4 and 6 pm is the
result of an afternoon class of 22 students held in this room. Clearly, the problem with this air
handling unit is that it provides the same ventilation rate of air to classes 120 as to classes of 20 or
even an empty room.

126
 ES 96 Spring 2002

9.15 Recommendations

9.15.1 Scheduling

A simple, cost-free solution to this problem is to utilize Apogee’s scheduling capabilities. Currently,
the system indexes the air handling unit to Occupied mode at 7 am and does not return to Unoccupied
mode until 7 pm. However, Apogee is capable of indexing the system to oscillate between Occupied
and Unoccupied mode at specified times in order to save energy by turning off heating, cooling and
electricity to fans. It would not be difficult then to program the system once at the beginning of each
semester in accordance with the class schedules for the semester.

Figure 9.10 presents a typical weekly class schedule

for G115. The schedule for Monday, Wednesday or
Friday, represents the classes producing the carbon
dioxide levels for Figure 9.9. The class of 117
students produces the large spike of 800 ppm while the
class of 22 students produces the smaller bump in the
afternoon.

As this graph suggests, the recommended schedule

solution would be wise because there are many hours
in the day that this classroom is not used. If a
temporary amendment needed to be added to the
schedule, however, it could easily be done by using
Apogee’s ability to add a preset “replacement day” or
time. Replacement days or times allow for the
scheduling of exceptions such as the extension school
class apparent at 7:30 pm on Tuesday.

Savings would thus result from turning the air

handling unit off during many hours of the day and
turning it on again only when the room will be

FIGURE 9.10 Class Schedule

127
 ES 96 Spring 2002

occupied. Judging by Figure 9.10, a conservative estimate may be made that the actual running time
of Air Handling Unit 1 could be decreased by 1/3.

Using the calculated annual cost of $6,300 from above, the new running cost for Air Handling Unit
would be $4,200 per year – a savings of $2,100 per year without incurring any cost. This savings
estimate is probably high since the operating costs are below the average cost when the room is
unoccupied because of the reduced heat load.

9.15.3 Reducing Ventilation

From calculations above:

‘Original’ operation supply rate of Air Handling Unit 1 for occupancy load = 3,380 cfm
Actual supply rate of Air Handling Unit 1 for occupancy load = 2,250 cfm

Total annual cost of running Air Handling Unit 1 would then be $4,200 per year33, a savings of
$2,100 per year — the same amount of savings found in the scheduler solution. However, costs for
changing the ventilation flow rate of an air handling unit may reach $15,000, as shown in 9.11.2, so
the costs for implementing this measure may not be worth the savings. Therefore, we recommend
enacting the cost-free scheduling solution.

9.15.3 Savings

In considering obtainable savings from our recommendation for Air Handling Unit 1, only
one contributor need be taken into account:
Reducing the total ventilation time during the weekdays corresponds to a reduction in annual
running costs

33
Based on the calculation of $1.86 per cfm per year

128
 ES 96 Spring 2002

In economic terms:

Recommendation Savings ($)

Adopt suggested scheduling system 2,10034
TOTAL 2,100

Table 9.8 Operation recommendations for AHU 1 and associated savings

9.15.4 Payback Period

Not applicable since the scheduling changes recommended would not require any installation fees or
additional costs. Minimal man power may be required to enact such changes once each semester,
possibly more often if “replacement days” are used. However, Harvard already has several
employees dedicated to managing the Apogee system, so this task would not require hiring any new
employees or significantly increasing the work load.

9.16 Findings for Air Handling Unit 3

9.16.1 Understanding Air Handling Unit 3

Air Handling Unit 3 is unique in that it is possesses a variable frequency drive functionality. When
indexed to operating mode, the air handling unit will measure indoor and outdoor enthalpy and if the
outdoor enthalpy is much higher than indoor enthalpy, dampers will be closed to their minimum
position to return as much indoor air as possible. If outdoor enthalpy level is acceptable, the system
will settle into the economizer program similar to Air Handling Unit 1. In the economizer program,
the variable frequency drive functionality takes a reading for the rate of air flow which is desired on
the ground floor, then depending on that rate, fans will expend more or less energy to provide a
higher or lower rate of air flow. The current setting of the operating sequence states that when the air
handling unit goes in to Occupied mode, it will stay there until 7 pm, at which point the system will
return to Unoccupied mode.

129
 ES 96 Spring 2002

9.16.2 Variable Air Volume Boxes

Once the air flow reaches the supply duct of the air handling unit, it is directed to variable air volume
boxes in each of the rooms. Variable air volume boxes, or VAV’s as they are commonly referred to,
are controlled by a thermostat placed in each room. Figure 9.11 shows the operating sequence of a
VAV box.

Figure 9.11 A typical VAV box in Maxwell Dworkin

Depending on whether the thermostat requests heating or cooling, the damper within the VAV box
will be positioned. Cooler air is supplied by the AHU, which tries to maintain a supply temperature
of 65 degrees. If the air needs to be heated, a heating coil located within the VAV box will do so. Air
is then supplied to the room.

When calculating the cost of conditioning and circulating one cfm per year, this additional heating
before entering the room had to be added to the heating provided by the air handling units when
necessary. The average temperature of the rooms supplied by VAV boxes was found, through
trended data, to be 72 degrees. Estimating alternative sources of heat such as electricity and body
heat to add slightly to the room’s temperature, our estimated supply temperature during the heating
months for each cfm of air was 70 degrees.

34
Based on a reduction in operating time of 1/3 and $1.86 annual cost per cfm

130
 ES 96 Spring 2002

Currently, the minimum and maximum damper positions for each VAV box are changeable values
written into the Apogee system code for each VAV box.

9.16.3 Actual Occupancy Load

Determining the minimum volume flow rate that may be provided by Air Handling Unit 3 is difficult
and might require tests run at different output flow rates or a study of the actual maximum occupancy
for these areas. However, a few things are clear:
At BOCA’S recommended volume flow rate for these areas, the current ventilation rate of
9,000 cfm is too high as it would provide adequate ventilate for 600 people on the ground
floor.
This number is also overestimating the volume flow rate of corridors to be 15 cfm per person.
At present, many VAV boxes are running at their minimum position, and unable to run any
lower, suggesting they may run lower if they could.

Rather than attempting to determine actual occupancy of the ground floor, we chose to focus our
solutions for savings on this last issue and the solution proposed by the office temperature models.

9.17 Recommendations

9.17.1 Temperature Model Solution

One source of savings for this air handling unit could be found by following the office temperature
model solution of increasing the deadband (offset) temperature range for each of the VAV boxes
supplying each room. Savings from such a solution are difficult to estimate since this is a variable
volume/variable frequency drive system that services nearly 20 VAV boxes on the ground floor of
Maxwell Dworkin.

131
 ES 96 Spring 2002

9.17.2 Reducing Ventilation

Changing the maximum and minimum allowable volume flow rates for each VAV box would also
prove to be a valuable source in savings since the cfm settings are currently set to numbers
determined by the ‘original’ maximum occupancy of each room. Savings from this solution would
also result from the reduction of energy required by the air handling unit fans to supply a lower
volume flow rate to the supply duct. Once again, savings are difficult to predict for a variable
volume/variable frequency drive system, but a conservative savings estimate may be made for a
possible reduction in the volume flow rate of 1,500 cfm.

From calculations above:

‘Original’ average supply rate of Air Handling Unit 3 for occupancy load = 9,000 cfm
Estimated supply rate of Air Handling Unit 3 for occupancy load = 7,500 cfm

Total annual cost of running Air Handling Unit 3 would then be $13,6500 per year35, a possible
savings of $2,730 per year. Changing the ventilation flow rate of Air Handling Unit 3 would incur no
costs since such changes may be achieved by a simple rewriting of the Apogee code.

9.17.3 Savings

In considering obtainable savings from our recommendation for Air Handling Unit 1, the following
two contributors are taken into account:

Reducing the constant ventilation rates of each VAV box corresponds to a reduction in
annual operating costs
Allowing a greater deadband for room air temperature provided by each VAV box to each
room corresponds to a reduction in cost of conditioning per cfm per year

35
Based on the calculation of $1.82 per cfm per year

132
 ES 96 Spring 2002

In economic terms:

Recommendation Savings ($)

Decrease ventilation requirements for each VAV box 2,73036
Allowing greater deadband for room air temperature in each room ?
TOTAL 2,730+
Table 9.9 Operation recommendations for AHU 3 and associated savings

9.17.4 Payback Period

Once again not applicable since the change in ventilation flow rates to each VAV box and the
changes in deadband temperatures would not require any installation fees or additional costs.
Minimal man power may be required to update the Apogee program and perhaps to analyze the
changes to determine their most efficient values.

36
Based on a reduction of 1,500 cfm at $1.82 annual cost per cfm

133
ES 96 Spring 2002

134
 ES 96 Spring 2002

10.0 LIGHTING IN MAXWELL DWORKIN

135
ES 96 Spring 2002

136
 ES 96 Spring 2002

10.1 Analysis of the Current Lighting Situation and Proposals to Improve

Efficiency

In addition to the office model, chilled water and Air Handling Units, our class also investigated the
opportunity for energy savings in lighting. Lighting accounts for forty percent of the electrical load in
Maxwell Dworkin. In order to create an energy-efficient, comfortable lighting atmosphere, an
important fact must be recognized: the human eye perceives relative lighting levels, not absolutes.
When we began our exploration of Maxwell Dworkin, it became clear to us that the building’s
lighting scheme did not take this significant fact into account. In essence, Maxwell Dworkin is being
unnecessarily overlit and increasing unnecessarily the air conditioning load.

10.1.1 Light Measurements and Units

Background:
Units of Measurement:
Before getting into the details of this problem, it is important to review basic lighting concepts. A
common unit of light is a lumen. A lumen measures the amount of actual light emitted from a light
source. However, a lumen is not a very practical unit for our class purpose. What we are interested
in is light levels in our working surfaces and spaces. A better unit to measure, therefore, is lux, which
is defined as one lumen per square meter of surface.

Recommended Light Levels:

It is also important to know that Massachusetts Building Code sets a maximum limit to the amount of
watts per square foot. However, watts per square feet does not acknowledge relative light levels in
the actual working space. It additionally provides minimum lighting requirements. In Section 780
CMR 1206.0, it states, “an average illumination of …64.58 lux over the area of the room at a height

137
 ES 96 Spring 2002

of 30 inches above the floor level” is required for “all habitable and occupiable rooms.”37 The
Illuminating Engineering Society38 (IES) examines this need by recommending a minimum light level
requirement. See Appendix 10 for IES tables. They suggest for an office space a level of 200 to 500
lux.

Measurements:
In order to get a feeling for this unit, it is useful to know some typical light measurements. Our class
used an EXTECH Instruments’ Lux meter as our tool. See Appendix 11 for EXTECH light meter
profile. We took measurements in a variety of locations in order to be able to compare relative light
levels. A dorm, at night with all the lights on, is around six hundred lux. The Fogg Museum light
levels range from fifty lux for hallways and walking areas and two hundred lux near the actual pieces
of art. In Maxwell Dworkin, on a sunny day, we measured average light levels on the Ground Floor
hall and the first, second and third floor hallways to be around three hundred lux. In a south-facing
office on a sunny day with the blinds up, the light levels reached eighteen hundred lux. In Pierce
Hall, the hallways have an average of 80 lux (Table 10.1). See Appendix 12 for complete data
measurement tables. Even within this small range of lighting measurements, the Maxwell Dworkin
office’s lighting level stands out. This result forced us to re-examine this room in closer detail.

TABLE 10.1 Measured Light Levels

Lux
Dorm Room 600
Fogg Museum 50-200
Maxwell Dworkin Hallway 300
Maxwell Dworkin Office 1800
Pierce Hall Hallway 80

Findings:
By just exploring Maxwell Dworkin with the light meter, we were able to pin-point two obvious
problems with the lighting scheme. It does not make use of daylight, or natural light, nor is it aware
of the activity of a location.

37
Massachusetts State Building Code User’s Guide to 780 CMR, Sixth Edition, published by William F.
Galvin, Secretary of the Commonwealth, February 1997.

138
 ES 96 Spring 2002

To investigate the wasting of natural light, our class evaluated a south-facing office. In the same
south facing office that reached 1800 lux, with only two thirds of the lights on and the blinds open,
the light levels remained sky high at 1700 lux. With all the lights on and half the blinds open, the light
levels reduced to 800 lux, which is still much higher than the value recommended by the Illuminating
Engineering Society. We can safely claim that sunlight streaming in through the windows provides at
least as much light as the electric light fixtures in the room. Not only was this office being overlit
from the inside, current lighting schemes do not take advantage of natural light. Lights are on when
there is sufficient day lighting to provide adequate levels for a work space.

Additionally, the lighting scheme does not pay attention to the activity in the location. Offices spaces
have their lights on when professors are not there. Lights in common spaces are on when they are
unoccupied. Additionally, the ground floor bathrooms remain on seven days a week, twenty four
hours a day. Surely, there are not people in them constantly. These two problems arouse after only a
brief inspection of Maxwell Dworkin. By continuing our investigation, more details emerged.

10.1.2 Light Bulb Types

Background:
Having gathered initial data concerning light levels in Maxwell Dworkin, we next explored the
various types of bulbs being used within the building. In order to understand the variance amongst
different types, it was important to become familiar with the main types.39

Incandescent Bulbs:
Incandescent bulbs possess a tungsten filament within a glass sphere. When electricity is passed
through the filament, it converts the electrical energy to heat. Once it gets hot enough, it emits white
light. These bulbs are highly inefficient. Halogen bulbs are simply slightly more efficient type of
incandescent bulb.

Fluorescent Bulbs:
Fluorescent bulbs are the most efficient light bulbs on the market. Basically, an electron stream
interacts with a gas, usually argon or mercury, within a glass tube. When the gas moves from the
excited to unexcited stage, they release photons which collide with the phosphorous coating on the

38
The Illuminating Engineering Society of North America website http://www.iesna.org/
39
A good website to visit is http://38.144.192.166/efficiency/lighting/index.html.

139
 ES 96 Spring 2002

tube, therefore producing visible light. Compact fluorescents, which are ideal for lighting specific or
smaller areas, are used mostly for down-lighting. T5 and T8 tube bulbs are the most common in
industrial spaces.

Description of the System:

For the most part, Maxwell Dworkin uses fluorescent bulbs, the most efficient bulbs on the market.
Offices and labs are outfitted with T8 fluorescent tube bulbs, which are a standard for efficiency in
work places. However, certain locales in the hallways contain incandescent down-lighting.
Additionally, lecture halls possess down-lighting also provided by incandescent bulbs. Halogen bulbs
are used for an aesthetic effect to highlight the walls outside of lecture hall G115. Maxwell Dworkin
is not always using the most efficient bulbs on the market.

10.1.3 Lighting Contrast Levels

Background:
The lighting scheme in Maxwell Dworkin does not take advantage of color contrast, an important
attribute of perception. As mentioned earlier, the human eye perceives relative light levels, not
absolutes. With appropriate contrast levels in a room, only 50 lux are needed40. Most people are not
even aware of the effect of contrast in their everyday lives. One example is a blackboard. It seems
obvious why white chalk is the best choice to use on a black board. If differences between
foreground and background colors are not distinct enough, contrast levels make it difficult to
distinguish between objects.

Description of the System:

Maxwell Dworkin does not take into account the importance of contrast in aiding vision. An example
of where Maxwell Dworkin’s contrast situation makes it difficult to see is in the hallways. Although
there is an adequate level of light striking the floor in the in the hallway, about 300 lux, the dark walls
and dark floor absorb the light being emitted by the light sources. Another example is the furniture in
Maxwell Dworkin. Most of the tables and desks, specifically in professors’ offices, are of light maple
wood. This wood is aesthetically pleasing, however it sets up a difficult contrast level between a
white piece of paper and the desk. By ignoring this basic concept of contrast, Maxwell Dworkin
requires more light to see than actually needed.

40
This information was obtained from Professor Michelle Addington of the Harvard Graduate School of
Design.

140
 ES 96 Spring 2002

10.1.4 Summary of Problems

In summary, our analysis of Maxwell Dworkin’s lighting scheme found that the building is overlit.
Specifically, it can be addressed through these four problems:
1. The lighting scheme does not take advantage of natural light.
2. It does not pay attention to the activity of a location.
3. It does not use the most efficient bulbs available.
4. It ignores the useful tool of contrast in its design.

Now that some obvious points for improvement in Maxwell Dworkin’s lighting scheme have been
introduced, potential solutions to these problems will be addressed.

10.2 Lighting Recommendations

All of the savings from modifying lighting come from two sources. First, running fewer lights means
using less electricity to operate the bulbs. Second, since each bulb gives off heat as it consumes
power, having fewer bulbs means the building will use less chilled water to remove the heat they give
off.

10.2.1 Ground Floor

On the ground floor we currently have, in use, two types of incandescent bulbs and halogen bulbs.
All of these types of bulbs are inefficient, so a large potential saving could be obtained from
switching them to dimmable compact fluorescent bulbs.

The bulbs we have now are 75 Watts, 100Watts, and 50 Watts. The dimmable compact fluorescents
which have the same lumen output run at 20 Watts, 28 Watts, and 15 Watts, respectively.

Savings:
$3003.86 per year41

41
See Appendix 13

141
 ES 96 Spring 2002

10.2.2 Faculty and Staff Offices

The offices for faculty and staff on the first, second, and third floors are overlit. Rather than replacing
the light mountings in order to optimize lighting, we have suggested a cheaper alternative – simply
removing some of the bulbs.

We recommend removing the middle of the three bulbs above the desk and above the door, and
removing the two outer bulbs in the other two light fixtures, as in shown Figure 10.1.

Switch 1
Switch 2

FIGURE 10.1 Proposed density of lighting in faculty and staff offices

This reduces the wattage used by the overhead lights by one half. We measured the lighting levels
with this configuration and they were still in a reasonable range.42 To compensate for lowering the
lighting levels, we suggest the purchase of a task lamp and a contrast pad for each desk. The task
lamp would be a fluorescent desk lamp which would be used to add illumination to the area on which
the resident of the office would be working. The contrast pad we used in our experimentation was a
simple, black compressed foam work pad. The purpose of this pad is to give the resident of the office
a dark area on their desk, over which it would be easier to read light colored papers.

We experimented with this arrangement in four faculty and staff offices over the course of two weeks,
and we received nothing but positive feedback.

Testimony:
Dr. Joy Sircar:
“I am completely satisfied with the new system. I believe the lighting meets my needs adequately. I
am enjoying the thought that I am saving some energy. Thank you for taking up the project.”

142
 ES 96 Spring 2002

Professor Barbara Grosz:

“The desk lamp is excellent. It's been mostly grey outside when I've been in my office, so that affects
my report on the new lighting levels. To have a meeting at the table I seem to need both the up and
down lights on, though it seems ok with a one-on-one meeting to have just the "downward" lights.
When I'm working at my desk then just having one set "downward" lights. When I'm working at my
desk then just having one set alone, please let me know. In any case having just the two sets working
and not the middle fixture seems quite fine.”

Peter Arvidson (graduate student):

“My contrast pad and low lighting is working fine.”

Costs:
45 offices x $52/task lamp = $2340
45 offices x $3/contrast pad = $135
Total: $2475

Savings:
Total: $3334 per year43

10.2.3 Hallways and Common Areas

Our final suggestion would be to install combination motion and light detectors in the hallways and
common areas. This addition would prevent the lights from being on when the area was unoccupied
or when there was sufficient outside sunlight to light the room.

The detector we used for our calculations is a Leviton ODS15-ID Occupancy Detector.44 Using its
specified range, we calculated that it would require 5 detectors to cover the basement walkways, 7 to
cover the ground floor lobby, hallways, computer station, and vending room, and 9 each on the first,
second, and third floors to cover the hallways and common areas. This comes to a total of 39
required detectors.

42
For a full list of measurements, see Appendix 13
43
See Appendix 13
44
For spec sheet, see Appendix 9

143
 ES 96 Spring 2002

Costs:
39 detectors x $60.95/detector = $2377.05
Installation: Unknown45

Savings:
Total: $16146.29 per year46

10.3 Summary of Savings

Ground floor bulbs: $3004 per year

Office lights: $3334 per year
Hallways/Common Areas: $16146.29 per year

Total: $22484 per year

45
We were not able to obtain quote for the installation costs of motion detectors. Every electrician we talked to
required an inspection of Maxwell Dworkin prior to making any estimate.
46
See Appendix 13

144
 ES 96 Spring 2002

11.0 GENERALIZATION TO OTHER

HARVARD BUILDINGS

145
ES 96 Spring 2002

146
 ES 96 Spring 2002

IMAGE 11.1 Science Center IMAGE 11.2 William James Hall

11.1 Applicability to William James Hall and the Science Center

11.1.1 Introduction

When we finalized our recommendations for Maxwell Dworkin, we felt that many of our findings

were applicable to other similarly operated Harvard University buildings. Although they emerged

from a detailed exploration of Maxwell Dworkin, these specific recommendations were developed

from more general findings that were not specific to Maxwell Dworkin.

In an attempt to determine whether or not these recommendations could be applied to other buildings,

we briefly examined both William James Hall and the Science Center. We identified ventilation and

lighting as two major areas that could potentially be improved in both buildings.

147
 ES 96 Spring 2002

11.1.2 Ventilation

In our exploration of the ventilation system we used the building’s CO2 levels to determine the
adequacy of the buildings ventilation rate. We applied the same test to both William James Hall and
the Science Center. We found that from a CO2 standpoint, both buildings were highly over
ventilated. In both buildings the CO2 level remained within a few hundred parts per million above
fresh air CO2 levels (400 ppm). In the Science Center the CO2 levels ranged between 450 and 575
ppm and William James Hall the levels fell between 475 and 650 ppm.

The AHU’s in William James Hall and the Science Center supply fresh air as well as heating and
cooling. Thus neither building has the fan coil units found in Maxwell Dworkin. Hence it seems fair
to predict that changes in these buildings’ HVAC systems may be simpler to implement than similar
changes in Maxwell Dworkin.

Although both buildings appeared to be over ventilated we did not make specific recommendations.
Our aim was to determine whether or not our general recommendations for Maxwell Dworkin were
applicable to other Harvard buildings. Thus we recommended a reduction in the two buildings’
ventilation levels.

11.1.3 Lighting

Both buildings also seemed to be over lit. The standard comfortable lighting level in public spaces

was specified by the Illuminating Engineering Society to be 200-500 lux. Our measurements revealed

that in many public spaces the lighting level was well above 500 lux. This suggested that William

James Hall and the Science Center were over-lit. We found that many of our lighting

recommendations for Maxwell Dworkin to be applicable to these two buildings. We recommended

the use of motion sensors, lux level meters, and color contrasts.

148
 ES 96 Spring 2002

11.1.4 Additional Savings

Brian Schoenbeck S.B. ‘02, developed a Windows-based software program that optimizes a the

economic efficiency of a buildings air conditioning system based on weather forecasts and local

conditions. This system would allow building operations managers to take advantage of energy

saving techniques (i.e. summer nighttime fresh air purges). Brian’s thesis project entitled A

Psychrometric Analysis System To Improve Air Conditioning Efficiency In Large Buildings, explains

his program in great detail. For further information contact the Harvard Division of Engineering and

Applied Sciences.

11.2 Conclusion

Before examining our energy saving recommendations for William James Hall and the Science

Center, the reader should consider two points. First, our recommendations do not represent the total

potential savings within these buildings. The energy saving recommendations only emerged form a

cursory look at these buildings and an attempt to apply generalized versions of the recommendations

applied to Maxwell Dworkin. Second, because both of these buildings are significantly larger than

Maxwell Dworkin, and both cost significantly more to operate (per square foot), it should be noted

that the magnitude of the savings recouped form the implementation of the energy savings

recommendations is potentially much greater than the potential savings for a building like Maxwell

Dworkin.

The savings recommendations for the ventilation and lighting systems within William James Hall and

the Science Center are listed below.

149
 ES 96 Spring 2002

Recommendations:

• Reduce ventilation levels

• Conduct night-time fresh-air purges

Reduce lighting levels,

• Install light level sensors & motion detectors

• Utilize color contrasts

150
 ES 96 Spring 2002

12.0 SUMMARY OF TOTAL SAVINGS

151
ES 96 Spring 2002

152
 ES 96 Spring 2002

12.1 Overall Savings

Based on the suggestions for improvements we have presented for Maxwell Dworkin, we may

calculate the potential improvement in energy efficiency. Crucial to understanding the environmental

savings are the conversion factors used to determine how many acres of US Forest are necessary to

sequester the amount of energy expended by chilled water, steam, and electrical usage within

Maxwell Dworkin. As previously calculated in Appendix 3, these conversions are as follows: 56 ton-

days of chilled water per acre; 12.5 MMBTU of steam per acre; 1350 kWh of electricity per acre.

To the list of modifications we add the temperature model’s proposed solution of allowing

temperature floating within the rooms. The savings in chilled water resulting from this measure

would amount to $14,000 per year. The 1,680 ton-days of chilled water this solution offers would

then be equal to the amount of carbon dioxide consumed by 30 acres of U.S. Forest. Other sources of

chilled water savings include the elimination of unnecessary weekend cooling, summer nighttime air

purges, and the installation of an ethylene glycol heat exchange system. These savings reach $17,000

per year or 1,850 ton-days per year sparing 33 acres of worth of carbon dioxide emission.

To avoid over-ventilating the building we suggest reducing the volume flow rate of air in the

building, utilizing scheduling capabilities, and reducing or floating the discharge temperature from air

handling units and VAV boxes. All three solutions provide savings in heating, cooling, and

electricity to run the fans. The total savings of $29,000 per year for all 5 air handling units would

account for a total environmental savings of 78 acres per year.

153
 ES 96 Spring 2002

Finally, we add electricity savings from installing motion detectors and changing the lighting

throughout the building. Total energy savings may possibly account for nearly $22,500. The 241.6

MWh of energy savings are the equivalent of 182 acres of US Forest.

These figures quickly add up to a possible savings of $82,500 for one year in just this one building.

The efficiency calculation of Maxwell Dworkin after these savings shows that it could potentially be

run at greater efficiency than even Pierce Hall—the building we determined had the greatest

efficiency of those we considered. Since Maxwell Dworkin’s initial operating budget was $316,680

per year for the 116,000 sq. foot building, the efficiency of the building was rated to be $2.73 per sq.

foot, well above Pierce Hall’s efficiency rating of $2.39 per sq. foot. Our solutions would reduce

Maxwell Dworkin’s operating budget to ~$234,000 per year, resulting in an efficiency rating of $2.02

per square foot. The monetary savings we have accounted for are certainly a motivating factor, but it

is truly the environmental savings that are priceless. The amount of energy saved would be equal to

the amount of carbon dioxide absorbed by nearly 323 acres of US forest.

In order to gain insight into the scale of the savings, 323 acres would equal 14 Harvard Yards—far

more expansive than the 220 acres of land that Harvard currently owns in Cambridge.

But not only are the possible savings impressive, the costs for implementing these measures are

minimal. An estimated one-time cost for buying motion detectors and task lamps would be $4,700,

though this does not include installation and assessment fees. The only further cost would be found

in the suggested improvements to Air Handling Units 4 and 5. The most basic solution involving

alternating the schedule of the Air Handling Units would involve a one-time cost of $4,000. The

alternative high-tech solution of installing a variable frequency drive on both of the Air Handling

Units as well as the Exhaust Fans would require an installation cost of $90,000. Therefore, enacting

all of our solutions and using the most low-tech solution for the Air Handling Units would amount to

a total one-time cost of $8,700 a sum that would pay for itself in just over one month assuming the

154
 ES 96 Spring 2002

full $82,500 was saved the first year. Under the same assumptions, enacting the more high-tech

solution involving installation of variable frequency drives would incur a cost of $94,700 which

would pay for itself in the first 14 months.

This project has shown the impressive savings that may be discovered by unleashing a group of

young engineers on a building that is built to be more efficient than many of Harvard’s buildings to

determine how that building may be run wiser. If this much may be saved on this one Harvard

building, imagine the results of applying our solutions to Harvard’s other buildings.

155
 ES 96 Spring 2002

APPENDIX 1

TEMPERATURE DATA FOR BOSTON, MA

ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 2

ENERGY BALANCE CALCULATIONS FOR

MAXWELL DWORKIN
ES 96 Spring 2002
 ES 96 Spring 2002

Note: . Constants are from textbook: Fundamentals of Thermodynamics by Sonntag, fifth edition, John Wiley
& Sons, Inc., 1998.
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 3

CONVERSION CALCULATIONS FROM

ENERGY SOURCE TO ACRES OF FOREST
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002

APPENDIX 4

UTILITY INFORMATION FOR

MAXWELL DWORKIN
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 5

CALCULATIONS AND MEASUREMENTS

FOR AHUS 4 AND 5
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 6

CALCULATIONS AND MEASUREMENTS

FOR AHUS 1 AND 3
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 7

SCHEMATIC DRAWINGS OF THE

OPERATIONS OF AHUS 1, 2, 3, 4 & 5
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 8

BOCA INTERNATIONAL MECHANICAL CODE

VENTILATION REQUIREMENTS
ES 96 Spring 2002
 SECTION 403
MECHANICAL VENTILATION

403.1 Ventilation system.

Mechanical ventilation shall be provided by a method of supply air and return or exhaust air. The
amount of supply air shall be approximately equal to the amount of return and exhaust air. The
system shall not be prohibited from producing negative or positive pressure. The system to
convey ventilation air shall be designed and installed in accordance with Chapter 6.
Ventilation supply systems shall be designed to deliver the required rate of supply air to the
occupied zone within an occupied space. The occupied zone shall have boundaries measured at 3
inches (76 mm) and 72 inches (1829 mm) above the floor and 24 inches (610 mm) from the
enclosing walls.

403.2 Outdoor air required.

The minimum ventilation rate of required outdoor air shall be determined in accordance with
Section 403.3.
403.2.1 Recirculation of air.
The air required by Section 403.3 shall not be recirculated. Air in excess of that required
by Section 403.3 shall not be prohibited from being recirculated as a component of
supply air to building spaces, except that:
1. Ventilation air shall not be recirculated from one dwelling to another or to
dissimilar occupancies.
2. Supply air to a swimming pool and associated deck areas shall not be recirculated
unless such air is dehumidified to maintain the relative humidity of the area at 60
percent or less. Air from this area shall not be recirculated to other spaces.
3. Where mechanical exhaust is required by Table 403.3, recirculation of air from
such spaces shall be prohibited. All air supplied to such spaces shall be
exhausted, including any air in excess of that required by Table 403.3.
403.2.2 Transfer air.
Except where recirculation from such spaces is prohibited by Table 403.3, air transferred
from occupied spaces is not prohibited from serving as makeup air for required exhaust
systems in such spaces as kitchens, baths, toilet rooms, elevators and smoking lounges.
The amount of transfer air and exhaust air shall be sufficient to provide the flow rates as
specified in Sections 403.3 and 403.3.1. The required outdoor air rates specified in Table
403.3 shall be introduced directly into such spaces or into the occupied spaces from
which air is transferred or a combination of both.

403.3 Ventilation rate.

Ventilation systems shall be designed to have the capacity to supply the minimum outdoor
airflow rate determined in accordance with Table 403.3 based on the occupancy of the space and
the occupant load or other parameter as stated therein. The occupant load utilized for design of
the ventilation system shall not be less than the number determined from the estimated maximum
occupant load rate indicated in Table 403.3. Ventilation rates for occupancies not represented in
Table 403.3 shall be determined by an approved engineering analysis. The ventilation system
shall be designed to supply the required rate of ventilation air continuously during the period the
building is occupied, except as otherwise stated in other provisions of the code.
 Exception:
The occupant load is not required to be determined, based on the estimated maximum
occupant load rate indicated in Table 403.3 where approved statistical data document the
accuracy of an alternate anticipated occupant density.
TABLE 403.3
REQUIRED OUTDOOR VENTILATION AIR
OCCUPANCY CLASSIFICATION ESTIMATED MAXIMUM OUTDOOR AIR
OCCUPANT LOAD, [cubic feet per minute (cfm)
PERSONS PER 1,000 per person] UNLESS
SQUARE FEETa NOTEDe
Correctional facilities
Cells 20 20
Dining halls 100 15
Guard stations 40 15
Dry cleaners, laundries
Coin-operated dry cleaner 20 15
Coin-operated laundries 20 15
Commercial dry cleaner 30 30
Commercial laundry 10 25
Storage, pick up 30 35
Education
Auditoriums 150 15
Classroom 50 15
Corridors — 0.10 cfm/ft.2
Laboratories 30 20
Libraries 20 15
Locker rooms —
0.50 cfm/ft.2
Music rooms 50
15
Smoking loungesb,g 70
60
Training shops 30
20
Food and beverage service
Bars, cocktail lounges 100 30
Cafeteria, fast food 100 20
Dining rooms 70 20
Kitchens (cooking)f,g 20 15
Hospitals, nursing and convalescent homes
Autopsy roomsb — 0.50 cfm/ft.2
Medical procedure rooms 20 15
Operating rooms 20 30
Patient rooms 10 25
Physical therapy 20 15
Recovery and ICU 20 15
Hotels, motels, resorts and dormitories
Assembly rooms
Bathroomsb,g 120 15
Bedrooms — 35 cfm per room
Conference rooms — 30 cfm per room
Dormitory sleeping areas 50 20
Gambling casinos 20 15
Living rooms 120 30
Lobbies — 30 cfm per room
30 15
Offices
Conference rooms 50 20
Office spaces 7 20
Reception areas 60 15
Telecommunication
centers and data entry 60 20
OCCUPANCY CLASSIFICATION ESTIMATED MAXIMUM OUTDOOR AIR
OCCUPANT LOAD, [cubic feet per minute (cfm)
PERSONS PER 1,000 per person] UNLESS
SQUARE FEETa NOTEDe
Private dwellings, single and multiple Based upon number of bedrooms. 0.35 air changes per houra or 15
Living areasc First bedroom: 2; each cfm per person, whichever is
additional bedroom: 1 greater
Kitchensg — 100 cfm intermittent or 25
cfm continuous
Toilet rooms and bathrooms g — Mechanical exhaust capacity
of 50 cfm intermittent or 20
cfm continuous
Garages, separate for each dwelling — 100 cfm per car
Garages, common for multiple unitsb — 1.5 cfm/ft.2
Public spaces
Corridors and utilities — 0.05 cfm/ft.2
Elevatorsg —
1.00 cfm/ft.2
Locker and dressing roomsb — 0.5 cfm/ft.2
— 75 cfm per water closet or
Toilet roomsb,g
urinal
Smoking loungesb,g 70 60
Retail stores, sales floors and showroom floors
Basement and street — 0.30 cfm/ft.2
Dressing rooms —
0.20 cfm/ft.2
Malls and arcades — 0.20 cfm/ft.2
Shipping and receiving —
0.15 cfm/ft.2
Smoking loungesb 70 60
Storage rooms — 0.15 cfm/ft.2
Upper floors — 0.20 cfm/ft.2
Warehouses —
0.05 cfm/ft.2
Specialty shops
Automotive service stations — 1.5 cfm/ft.2
Barber 25 15
Beauty 25 25
Clothiers, furniture — 0.30 cfm/ft.2
Florists 8 15
Hardware, drugs, fabrics 8 15
Pet shops — 1.00 cfm/ft.2
Reducing salons 20 15
Supermarkets 8 15
OCCUPANCY CLASSIFICATION ESTIMATED MAXIMUM OUTDOOR AIR
OCCUPANT LOAD, [cubic feet per minute (cfm)
PERSONS PER 1,000 per person] UNLESS
SQUARE FEETa NOTEDe
Theaters
Auditoriums 150 15
Lobbies 150 20
Stages, studios 70 15
Ticket booths 60 20
Transportation
Platforms 100 15
Vehicles 150 15
Waiting rooms 100 15
Workrooms
Bank vaults 5 15
Darkrooms — 0.50 cfm/ft.2
Duplicating, printing — 0.50 cfm/ft.2
Meat processingc 10 15
Pharmacy 20 15
Photo studios 10 15
Sports and amusement
Ballrooms and discos 100 25
Bowling alleys (seating areas) 70 25
Game rooms 70 25
Ice arenas — 0.50 cfm/ft.2
Playing floors (gymnasiums) 30 20
Spectator areas 150 15
Swimming pools (pool and deck area) — 0.50 cfm/ft.2
Storage
Repair garages, enclosed public garagesd — 1.5 cfm/ft.2
Warehouses — 0.05 cfm/ft2

For SI: 1 cubic foot per minute = 0.0004719 m3/s,

1 cubic foot per minute per square foot = 0.00508 m3/(s ◊ m2),
°C. = [(°F.) -32]/1.8.
a Based upon net floor area.
b. Mechanical exhaust required and the recirculation of air from such space as permitted by Section
403.2.1 is prohibited (see Section 403.2.1).
c Spaces unheated or maintained below 50°F. are not covered by these requirements unless the occupancy
is continuous.
d. Ventilation systems in enclosed parking garages shall comply with Section 404. A mechanical
ventilation system shall not be required in garages having a floor area not exceeding 850 square feet and
used for the storage of not more than four vehicles or trucks of 1 ton maximum capacity.
e Where the ventilation rate is expressed in cfm/ft.2, such rate is based upon cubic feet per minute per
square foot of the floor area being ventilated.
f The sum of the outdoor and transfer air from adjacent spaces shall be sufficient to provide an exhaust
rate of not less than 1.5 cfm/ft2.
g Transfer air permitted in accordance with Section 403.2.2.
 ES 96 Spring 2002

APPENDIX 9

DECORA WALL SWITCH OCCUPANCY SENSOR

ES 96 Spring 2002
 Decora Wall Switch ODS15-ID
Occupancy Sensor

Features Self-Adjusting Technology with

Automatic “Walk-Through’ Sensing
APPLICATION
Leviton’s Cat. No. ODS15-ID Decora Wall Switch Passive Infrared
(PIR) Occupancy Sensor is used to provide automatic lighting con-
trol for energy savings and convenience in a variety of commercial
applications, including:
• Small offices • Conference rooms • Storage areas
• Restrooms • Class rooms • Lounges
The ODS15-ID can be used for automatic switching of incandescent
lamps and low-voltage lighting with electronic and magnetic ballasts.
The unit also features a manual override switch that can be used to
keep lights OFF while an area is occupied, which may be desired in
conference rooms and other areas during slide or film presenta-
tions. The unit installs in place of a single-pole wall switch and fits in
a standard wall box. The unit requires a ground connection.
OPERATION
The ODS15-ID uses passive infrared (PIR) detection technology to
monitor a room for occupancy through a segmented Fresnel lens.
This specialized lens divides the field of view into sensor zones.
When a person passes into or out of a sensor zone, the sensor
detects motion and switches the lights ON. The lights will remain ON
as long as there is an occupant moving through the sensor zones.
Cat. No. ODS15-ID
A delayed-OFF time adjustment prevents the lights from switching
OFF when the space is occupied. In order to keep the lights ON, a
Push-button Manual Override Control
person must pass through a sensor zone at least once during the
For manual control, the ODS15-ID features a convenient push-button
selected delayed-OFF time interval. An LED indicator blinks each time
switch. If the lights are OFF, pressing the button will turn lights ON
the unit detects activity in the sensor zones. When the space being
and keep them ON for as long as the room is occupied. The lights
monitored by the sensor is unoccupied for the length of time chosen
will be turned OFF once the room is vacant, after the delayed-OFF
as the delayed-OFF interval, the unit will beep 3 times. Ten seconds

after the last warning beep, the unit will switch the lights OFF.
To ensure longer service life and compatibility with electronic ballasts,
the device carefully times its switching contact opening and closing
with the zero crossing point of the AC power curve. This minimizes
contact wear caused by in-rush currents from electronic ballasts.
An exclusive Walk-Through feature addresses the common
situation where personnel may only enter a room momentarily. The
Walk-Through feature provides increased energy savings by not leav-
ing the lights ON for an extended period after only momentary occu-
pancy. The unit will switch lights ON when it detects a person
entering the area it is monitoring. However, if the ODS15-ID does not
continue to detect activity during the 2-1/2 minutes immediately fol-
lowing the initial entry, it will automatically go to a short 2-1/2 minute
delayed-OFF mode. After this 2-1/2 minute time interval expires, the
unit will beep 3 times and then wait 10 seconds before switching the
lights OFF.
ODS15-ID
time expires. If the lights are ON, pressing the button will turn lights
OFF and keep them OFF even if the room is occupied. This feature is
particularly useful for slide or film presentations. The lights can be
turned back ON by simply pressing the button. The unit will then
return to normal operation. If the button is not pressed to turn the
lights back ON and the unit does not detect any motion during the
delayed-OFF time interval, the lights will remain OFF. The unit will
then return to normal operation where the lights will remain OFF
until it detects occupancy and automatically switches lights ON.
Manual-ON/Auto-OFF Mode
In this mode, the unit will not turn lights ON automatically when
motion is detected. Lights can only be turned ON manually by
pressing the push-button. The lights will remain ON as long as the
unit detects activity in the sensor zones. It will shut lights OFF auto-
matically after the space becomes unoccupied and the delayed-OFF
time expires. Lights can also be turned OFF manually at any time by
pressing the push-button. This mode is ideal for areas where man-
ual ON switching is required but automatic OFF switching is desired
for energy savings.

LIGHTING CONTROL DIVISION

Leviton Mfg. Co., Inc. 59-25 Little Neck Pkwy • Little Neck, NY 11362-2591 • Tech Line: 1-800-824-3005 • Fax: 1-800-832-9538

M21
 Decora Wall Switch
Occupancy Sensor ODS15-ID
.........................
FIELD OF VIEW
Top View
The ODS15-ID provides a180° field of view with a maximum cover-
age area of approximately 2100 square feet. The maximum sensing
distance in front of the sensor is 40 feet, and at each side is 30 feet.
A “small-motion” zone detects relatively small body movements and
allows the lights to stay ON even though a person may not be mov-

40 Feet Large-Motion Zone

ing or walking around the room. The remainder of the field of view,
the “large-motion” zone, exhibits a lesser degree of sensitivity and
requires larger movements.

Small-Motion Zone
20 Feet
30 Feet
Small-Motion Zone
60 Feet Large-Motion Zone

ENHANCED ADJUSTMENT OPTIONS Range

The ODS15-ID is preset to deliver optimum performance in a wide Reducing the coverage range allows the unit to ignore motion at the
variety of commercial applications. There are optional adjustments far end of its range and avoid unnecessarily switching lights ON. The
for sensitivity, ambient light override, delayed-OFF time, and field- range can be adjusted from 100% to 36% of the total coverage area.
of-view. These adjustments will customize the performance to meet Light
the needs of a specific installation. To avoid tampering, all adjust- The ODS15-ID is factory preset without any ambient light override in
ments can only be accessed by removing the control panel cover. A effect. This means the unit will switch lights ON when it detects
small flat-head screwdriver can be used to adjust the control knobs, occupancy, regardless of the amount of natural sunlight present. To
and the field-of-view blinders are finger-tip operated. Controls are maximize energy savings in some installations, the ambient light
labeled as follows: override feature will prevent the unit from switching lights ON when
Blinders there is ample natural sunlight, regardless of occupancy. This adjust-
Integral sliding blinders on each side of the lens may be used to ment should be made when the ambient light is at the level where
restrict the 180° field of view down to 32°. This will prevent no artificial light is needed.
unwanted detection in areas such as hallways. Self-Adjusting Ambient Light Override Feature—After the
ambient light override level has been set, the sensor will compen-
Time
sate for changes in lighting requirements. The push-button can be
The delayed-OFF time is preset at 10 minutes. A choice of four
used to turn lights ON at any time, regardless of the override setting.
delayed-OFF time settings is available: 30-seconds (for walking test
When the push-button is used, the unit automatically resets the
purposes only), 5 minutes, 10 minutes, and 20 minutes. Unit will
ambient light override based on the natural light level at that time.
beep when adjusting knob is set to the indicated time value.
To return the ambient light override back to the original setting,
Self-Adjusting Delayed-OFF Time Feature—Provides press and hold the push button down for approximately three sec-
optimum convenience and energy savings by preventing unneces- onds until the unit beeps, indicating the original level is restored.
sary ON/OFF light switching. During periods of occupancy where
there is little activity and infrequent motion detection, the delayed-
OFF time setting is automatically increased, up to a maximum of
30 minutes. During periods of occupancy where there’s more
activity and frequent motion detection, the delayed-OFF time set-
ting will automatically be decreased. The decreased delayed-OFF
time setting will be in effect after the unit has timed out and the
next period of occupancy begins. The delayed-OFF time will not be
reduced beyond the preset 10 minute interval or the custom-
selected time interval.

LIGHTING CONTROL DIVISION

Leviton Mfg. Co., Inc. 59-25 Little Neck Pkwy • Little Neck, NY 11362-2591 • Tech Line: 1-800-824-3005 • Fax: 1-800-832-9538

M22
 Decora Wall Switch
Occupancy Sensor
SPECIFICATIONS:
The device listed herein shall be Leviton Commercial Specification
Grade Decora Wall Switch Occupancy Sensor, capable of detecting
infrared emissions from human presence and responding by switch-
ing incandescent, low-voltage, and fluorescent lighting loads on. If
this unit does not detect movement after a present period of time, it
will respond by switching its assigned load off. The unit shall switch
at the zero crossing point of the AC power curve to ensure maxi-
mum relay contact life and compatibility with electronic ballasts.
Wall Switch Occupancy Sensor shall be equipped with a push-
button to provide manual on/off switching. Unit shall provide a
walk-through feature to prevent lights from remaining on for
extended periods after momentary occupancy in the area being
monitored. Leviton Decora Wall Switch Occupancy Sensor shall
feature self-adjusting delayed-OFF time and ambient light override
capabilities. Unit shall also provide sensitivity adjustment and inte-
gral sliding blinders to customize the horizontal field of view. Unit
shall be capable of providing optional manual-on/automatic-off
operation.
FEATURES AND BENEFITS
• Self-adjusting delayed-OFF time interval compensates for real-time
occupancy patterns, preventing unnecessary ON/OFF switching.
• Exclusive Walk-Through feature provides increased energy savings
by not leaving the lights ON for an extended period after only
momentary occupancy.
• New, low-profile design eliminates obtrusive “scanning-device”
look. Elegant Decora styling complements any interior; uses
Decora wallplates and coordinates with Leviton’s popular line of
Decora wiring devices.
• 180° field-of-view provides approximately 2100 square feet of
coverage suitable for small offices, conference rooms, class rooms,
stock rooms, lounges, restrooms, warehouses and a variety of
commercial areas.
ODS15-ID
.........................
• Convenient push-button provides manual ON/OFF light switching
at any time.
• Segmented Fresnel lens provides optimum sensitivity and perfor-
mance. Designed with an extensive “small motion” area where
even slight body movements will be detected.
• Horizontal field of view may be adjusted between 180° and 32° of
arc by using integral blinders located on either side of the lens.
• Optional manual adjustment for delayed-OFF time settings of 30
seconds (for walking test), 5 minutes, 10 minutes and 20 min-
utes. Allows customized adjustments to maximize energy savings.
• Adjustable Ambient Light Override ranges from approxi-mately 2
foot-candles (2 lux) to 500+ foot-candles (500+lux) to prevent
lights from turning ON automatically during periods of ample nat-
ural light, increasing energy savings.
• Self-Adjusting Ambient Light Override compensates for changes in
lighting requirements after override level has been set.
• Manual-ON/Automatic-OFF mode for installations where manual
ON switching is required but automatic OFF switching is still
desired for energy savings.
• LED indicator light flashes when sensor detects motion to verify
detection is active.
• Unit beeps 3 times after delayed-OFF time expires, then waits 10
seconds before turning lights OFF.
• One unit can be used for either 120V or 277V lighting. Compatible
with both electronic and magnetic ballasts.
• Relay switches at the zero crossing point of the AC power curve
to ensure maximum contact life and compatibility with electronic
ballasts.
• Fits in standard wallbox and replaces single-pole wall switch;
ground connection required. Gangable with other units.
• UL Listed and CSA Certified, complies with California Title 24
Energy Code and FCC Regulations.
• Limited Five-Year Warrant

DIMENSIONAL DIAGRAMS

1.75
(44.4 mm) Mounting Yoke
1.30
0.48 (12.6 mm) Lens
(33.0 mm)

1.35
(34.4 mm)

Control Panel,
Cover Removed
BLINDERS
4.06 (103.2 mm)
2.61 (66.3 mm)

2.60 (66.2 mm)

10 20 4 7 4 7

30 10 10
TIME RANGE LIGHT
Led Window

Push Button

OFF ON OFF ON
AUTO AUTO
BOTTOM
BOTTOM Service Switch

Cat. No. ODS15-ID

LIGHTING CONTROL DIVISION

Leviton Mfg. Co., Inc. 59-25 Little Neck Pkwy • Little Neck, NY 11362-2591 • Tech Line: 1-800-824-3005 • Fax: 1-800-832-9538

M23
 Decora Wall Switch
Occupancy Sensor ODS15-ID
.........................
INSTALLATION WIRING DIAGRAM
The ODS15-ID is preset to deliver optimum performance in a Black
wide variety of applications without requiring any adjustments To Line
during installation. Exclusive self-adjusting operating features will
automatically compensate for real-time occupancy patterns to
provide maximum convenience and energy savings. The unit
may replace a single-pole wall switch mounted in a standard
wallbox. The unit must be properly grounded in order to oper- Blue White
To Neutral
ate. The unit’s integral blinders may be used to restrict the field
of view to prevent unwanted detection of hallway traffic. It
should be positioned at least 4 feet away from HVAC registers. Green
The switch located under the push-button provides three
Controlled
operational setting: “ON” turns the lighting load ON indefinitely Load
(motion detection not active), “OFF” disconnects the lighting
load from the power for relamping purposes, and “AUTO”
activates automatic motion detection. Note that whenever the
unit is powered up, it will take approximately 1 minute to begin
normal operation. To Ground

Cat. No. ODS15-ID

PHYSICAL SPECIFICATIONS

Operating Temperature Range 0°C to 50°C

Storage Temperature Range -10°C to 85°C
Relative Humidity 20% to 90% non-condensing
Agency Approval UL Listed/CSA Certified
Complies with California Title 24
Energy Code
Complies with FCC Regulations

ELECTRICAL REQUIREMENTS
Line Voltage 120/277 VAC
Operational Frequencies 60Hz
Wire Designation Line—Black
Load —Blue
Ground—Green
Load Rating Fluorescent:
1800VA @ 120V
4000VA @ 277V
Incandescent:
1800W @ 120V
Motor: 1/4 HP @ 120V

ORDERING INFORMATION
Cat. No. Cat. No. Cat. No. Cat. No. Description
Ivory White Gray Almond
ODS15-IDI ODS15-IDW ODS15-IDG ODS15-IDA Decora Wall Switch
Occupancy Sensor,
with self-adjusting features
120/277V rating

Visit our Website at: www.leviton.com

LIGHTING CONTROL DIVISION
Leviton Mfg. Co., Inc. 59-25 Little Neck Pkwy • Little Neck, NY 11362-2591 • Tech Line: 1-800-824-3005 • Fax: 1-800-832-9538

M24 Copyright ©2001 Leviton Lighting Control Division. All rights reserved.
Subject to change without notice G-5520A/A1-DP
 ES 96 Spring 2002

APPENDIX 10

LIGHTING TABLES FROM THE ILLUMINATING

ENGINEERING SOCIETY OF NORTH AMERICA
ES 96 Spring 2002
 ES 96 Spring 2002

Excerpt from http://www.mts.net/~william5/library/illum.htm#3.

Footcandles and Lux for Architectural Lighting
(An introduction to Illuminance)

(c) 1999, Bill Williams, Edition 2.1

4.) IES - RECOMMENDATIONS

Since 1958 the Illuminating Engineering Society has published illuminance

recommendations in table form. These tables cover both generic tasks (reading, writing etc),
and 100's of very specific tasks and activities (such as drafting, parking, milking cows,
blowing glass and baking bread).

All tasks fall into 1 of 9 illuminance categories, covering from 20 to 20,000 lux, (2 to 2000
footcandles). The categories are known as A - I, and each provide a range of 3 iluminance
values (low, mid and high). See Table 1.

=====================================================================
TABLE 1
=====================================================================

IES ILLUMINANCE CATEGORIES and VALUES - for GENERIC INDOOR ACTIVITIES

ACTIVITY CATEGORY LUX FOOTCANDLES

----------------------------------------------------------------------
Public spaces with dark A 20-30-50 2-3-5
surroundings
----------------------------------------------------------------------
Simple orientation for short B 50-75-100 5-7.5-10
temporary visits
----------------------------------------------------------------------
Working spaces where visual C 100-150-200 10-15-20
tasks are only occasionally
performed
----------------------------------------------------------------------
Performance of visual tasks of D 200-300-500 20-30-50
high contrast or large size
----------------------------------------------------------------------
Performance of visual tasks of E 500-750-1000 50-75-100
medium contrast or small size
----------------------------------------------------------------------
Performance of visual tasks of F 1000-1500-2000 100-150-200
low contrast or very sm size
----------------------------------------------------------------------
Performance of visual tasks of G 2000-3000-5000 200-300-500
low contrast or very sm size
over a prolonged period
----------------------------------------------------------------------
Performance of very prolonged H 5000-7500-10000 500-750-1000
and exacting visual tasks
 ES 96 Spring 2002

----------------------------------------------------------------------
Performance of very special I 10000-15000-20000 1000-1500-2000
visual tasks of extremely low
contrast
----------------------------------------------------------------------
A-C for illuminances over a large area (ie lobby space)
D-F for localized tasks
G-I for extremely difficult visual tasks
 ES 96 Spring 2002

APPENDIX 11

EXTECH LIGHT-METER PROFILE

ES 96 Spring 2002
ISO 9001 CERTIFIED
NEWPRODUCT
Heavy Duty Light Meter
Simultaneous display of Foot Candles or Lux plus four lighting types —
Now includes Sodium light measurements

Features:
• Microprocessor assures maximum accuracy plus
special functions
• Super large 1.4” (1999 count) LCD display
• % displays differential from reference point
• “ZERO” Re-Calibration
• Utilizes precision photo diode and color correction filter
• Cosine and color corrected measurements
• ±4% accuracy is enhanced by selecting lighting Optional Data Acquisition
type (Tungsten/Daylight, Fluorescent, Sodium, Mercury) Software and Datalogger for
Later Analysis
• Record/Recall MIN, MAX and AVG readings
• Data Hold plus Auto shut off
• Built-in RS232 serial interface, optional software
• Dimensions 7x2.9x1.3” (178x74x33mm); weight
1.6lbs. (726g) Measuring Ranges:
• Complete with 9V battery and protective holster. Fc LUX
200.0 2000
2000 20000
Interface Options: 5000 50000
• Optional Windows™ Data Acquisition Software
and serial cable enable user to display and capture
readings on a PC, and set time intervals and alarms
• Optional battery operated Datalogger module stores
over 8000 readings for later transfer to a PC

Ordering Information:
407026..................Foot Candle/Lux Meter
407026-NIST ......Foot Candle/Lux Meter with NIST certificate
407001..................Data Aquisition Software and Serial Cable
380340..................Datalogger Module with PC-DOS software
156119 ................117 VAC Adaptor
156221..................220 VAC Adaptor
409997 ................Soft Vinyl Pouch Carrying Case
140001..................Hard vinyl carrying case with handle

EXTECH INSTRUMENTS CORPORATION

285 Bear Hill Road, Waltham, MA 02451-1064
Phone: 781-890-7440 • FAX 781-890-7864
www.extech.com
The Source for ALL
Your Measurement Needs!
3 Year
Warranty
02/02 - R1
 ES 96 Spring 2002

APPENDIX 12

LIGHTING MEASUREMENTS FOR A SOUTH-

FACING OFFICE SPACE ON A SUNNY DAY
ES 96 Spring 2002
ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 13

LIGHTING MEASUREMENTS AND CALCULATIONS

ES 96 Spring 2002
ES 96 Spring 2002
 ES 96 Spring 2002

APPENDIX 14

TELAIRE® 7001 CARBON DIOXIDE AND

TEMPERATURE MONITOR
ES 96 Spring 2002