Professional Documents
Culture Documents
)
In Equation 1, b indexes all building types in the demand model, t indexes time and is bounded
from 1 to 8,760. Q, C, and E are the heating, cooling, and electrical loads (J/m
2
) in each hour for
each building, respectively. v
b
denotes the floor area of each building type. LHV is the lower
heating value (J) of natural gas consumed in each hour.
Optimization. Optimization was performed in ModelCenter, a commercial software package
from Phoenix Integration that allows integration of simulation programs via plug-ins and
software wrappers written in the Python programming language (PHX, 2008). Decision variables
are chosen in ModelCenter and used to control linked simulation files (e.g., Microsoft Excel) for
calculation of the objective function. This process is repeated for each candidate solution until
an optimum is found. For this case study, a single objective of maximizing TFCE was used (see
Equation 1). The SEQOPT sequential quadratic programming algorithm was used to find a
solution. This algorithm statistically samples the potential solution space using an orthogonal
array and then uses quadratic programming on simplified surrogate models and pattern search
to iteratively improve the local optima (Booker, et al., 1999). The algorithm ceases when
incremental improvement in the objective function is lower than 10
-5
.
The decision variables, v
b
, were 26 continuous parameters describing the gross floor area of
each building type. A dependent integer variable ranged between 1 and 5 as a multiplier for the
number of gas engines in the CCHP plant, dictating its size. A constraint was placed on the total
electrical demand, E
max
, such that it could not exceed 47.5 MWh, the capacity of the 5 gas
engines. Constraints were also placed on the maximum percentage of each of building type, V
b
,
so that no single building could exceed 25% of the total built-up area. These ensure a more
realistic scenario by varying heights and uses. Non-negativity constraints on each building were
also included. The constraints are shown in Equations 2 and 3.
Best, et al.
If applicable, page number will go here after aggregating all papers
Results, Analysis, and Validation. The optimization program was run on a computer with two
Quad Core Xeon E5440 processors (2.83 GHz) and 16 GB of RAM. The maximum efficiency
optimization ran to completion in 3,951 iterations in 8.5 hours; 7 similar trials were run to
generate additional solutions and find the minimum efficiency. The mix of uses in the CBD and
City were specified in the Integrated Energy Model and the TFCE values recorded without
optimization. A second optimization was performed to minimize TFCE to characterize the
feasible solution space. Table 2 shows the TFCE and hourly standard deviation of all four
scenarios. The average fraction of electrical and heat demand from the gas engines is also
shown. In the CBD, City, and Maximum Efficiency scenarios, five combustion engines were
chosen by the simulation for an electrical capacity of 47.5 MW; the Minimum Efficiency scenario
used two combustion engines, for a capacity of 19 MW. An ideal scenario where all energy from
the CCHP plant is consumed is also shown for comparison.
Table 2: Characteristics of the City, CBD, Maximum Efficiency, and Minimum Efficiency scenarios.
Run Total Fuel Cycle
Efficiency
Hourly Standard
Deviation in TFCE
Average Electrical
Fraction
Average Heat
Fraction
Ideal Scenario 87.5% 0.00% 0.55 0.45
Maximum Efficiency 55.67% 6.95% 0.65 0.35
City Baseline 45.33% 9.91% 0.79 0.21
CBD Baseline 44.92% 9.46% 0.76 0.24
Minimum Efficiency 35.77% 7.93% 0.67 0.33
Figure 2 shows the mix of building uses for each scenario. Uses are characterized into 8
primary use types by aggregating similar uses of different forms present in the 26 prototype
buildings.
Figure 2: Results of Optimization. Chart showing the mix of uses for the Maximum Efficiency, Minimum Efficiency,
City, and CBD Scenarios. The Maximum Efficiency scenario shows significantly higher proportions of medical,
lodging, and retail buildings than the City and CBD. The Minimum Efficiency has a large amount of residential space.
The results of the simulation and optimization indicate the potential to increase TFCE 10% and
12% from the City and CBD scenarios, respectively. Examination of the building types in the
Maximum Efficiency scenario reveals that the efficiency increase stems from an unviable, large
percentage of medical, retail, and hospitality buildings. The hourly profiles of these types contain
temporally similar space conditioning and electrical load profiles, creating a more constant
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Maximum
Efficiency
Minimum
Efficiency
City Scenario CBD Scenario
Mix of Building Uses for Simulated Scenarios
Retail
Residential
Office
Medical
Lodging
Industrial
Education
Optimizing the Urban Plan of a CCHP-Powered Community for Total Fuel Cycle Efficiency
If applicable, page number will go here after aggregating all papers
utilization of electricity and heat from the gas engine throughout the year. The baseline studies,
by comparison, have a larger share of office and residential buildings, which contribute to a
coincident afternoon electrical peak without corresponding heating or cooling. This is also
reflected in the average heat and electricity fractions shown in Table 2; more equal heat and
electricity fractions exist in the Maximum Efficiency scenario. These fractions are closer to the
ideal as well, indicating more complete use of the heat and electricity generated from natural
gas combustion. Furthermore, the reduced standard deviation in hourly TFCE values indicates
reduced numbers of extreme lows and highs in efficiency, and greater clustering of hourly TFCE
values around the higher mean. This is partially a result of the more constant loads generated
by the mix of uses chosen in the optimal scenario; such clustering can also be in part an artifact
of the optimization.
It is worth noting that the maximum efficiency scenario is still far short of meeting the ideal
utilization of heat, cooling, and electricity, indicating that for this scenario, it is impossible to
balance perfectly the output characteristics of the CCHP plant by only varying the mix of
building uses. A gap still exists in many hours between the thermal and electrical demand and
supply; additional schedulable electrical and thermal uses not tied to the mix of buildings or
load-shifting technologies may help balance the supply and demand of electricity, heating, and
cooling. However a study of these potential solutions is beyond the scope of this work.
To validate the results, a Monte Carlo analysis (MCA) was performed incorporating three main
sources of variability. The first is the accuracy of the prototype building hourly energy results as
representations of the actual buildings. Fumo, et al., among others, have characterized the
standard deviation of the mean of this uncertainty as 10% annually (2010). The distribution is
assumed to be normal by the Central Limit Theorem given that the error results from hundreds
of choices, each with unique distributions. The second is the hourly forecast error inherent in
using energy modeling as a predictive tool for planning electricity, heating, and cooling
production. Sevlian and Rajagopal have shown that for clusters of 200 buildings or greater, the
standard deviation of the mean is stable at 10%, and has a normal distribution (2013). Sevlian
and Rajagopal show this for electricity; it is assumed to be true for heating and cooling as well.
The final error is in efficiency of the CCHP engine at a given operating point. This was deduced
from documentation from General Electric to be normal with a standard deviation of 0.75 at any
point along the efficiency curve (Payrhuber, K. and Trapp, C., 2011).
MCA was performed on all four scenarios. 30 trials were simulated for each drawing from all
three error distributions. The resulting distributions are shown in Figure 3. The distributions
show no overlap between the maximum and minimum efficiency scenarios and either baseline.
Furthermore, a t-test shows that the minimum and maximum efficiency cases are different from
each other and from both baselines with greater than 99% confidence. The two baselines
cannot be shown to be unique from one another.
Best, et al.
If applicable, page number will go here after aggregating all papers
Figure 3: Probability Density Functions from MCA of results. The distributions above were generated from a
Monte Carlo Analysis of the Total Fuel Cycle Efficiency of all four scenarios. No perceptible overlap is evident,
indicating that the maximum efficiency scenario is distinct from both baselines and the minimum efficiency case.
Conclusions and Future Work. A retrospective case study of repowering a scaled model of
Oakland, CA, with a 47.5 MW CCHP plant demonstrated that optimization of the mix of building
uses can improve Total Fuel Cycle Efficiency up to 12%. This increase results from a higher
percentage of uses with similar electricity and space conditioning hourly load profiles. This study
assumed that the hourly heating, cooling, and electrical demand of the community could be
estimated by aggregating the 16 EnergyPlus Reference Prototypes and 10 additional models.
Additional research is required to fully validate the accuracy of this approach for Oakland, CA,
and for this type of modeling. Additional research should also explore the implications of using
more prototype buildings that provide a more granular set of uses and densities. Specific
technology was selected for the CCHP plant as well; exploration of the solution space with
different equipment and control schemes should be explored. In addition, this study used a
simplified method to calculate heating, cooling, and electrical losses based solely on the
distance of the building from the CCHP plant. Further work is planned to model the layout and
corresponding losses of the distribution network in more detail.
Future research will focus on extending simulation to other sources of energy and other uses to
allow simultaneous optimization of supply and demand that may better balance the thermal and
electrical demand and supply. It is also recognized that the mix of uses in the maximum
efficiency scenario presented here is not socially viable; we plan to consider additional social or
economic objectives and constraints in the future to better reflect the variety of considerations
urban planners face when designing new communities.
Acknowledgements. The authors would like to thank the Center for Integrated Facility
Engineering at Stanford University, Walt Disney Imagineering and Dr. Ben Schwegler, and the
National Science Foundation Graduate Research Fellowship Program for supporting this work.
0
50
100
150
200
250
300
350
400
0.3 0.35 0.4 0.45 0.5 0.55 0.6
Probability Density Functions for Efficiencies of All
Four Scenarios
Max Efficiency Min Efficiency City CBD
Optimizing the Urban Plan of a CCHP-Powered Community for Total Fuel Cycle Efficiency
If applicable, page number will go here after aggregating all papers
References
Akisawa, A., Kang, Y., Shimazaki, Y., & Kashiwagi, T. (1999). Environmentally Friendly Energy
System Models Using Material Circulation and Energy Cascade--The Optimization Work.
Energy, 24: 561-578. DOI: http://dx.doi.org/10.1016/S0360-5442(99)00023-7.
Boehm, B. (2010). Experimental Determination of Heat Losses from Buried District Heating
Pipes in Normal Operation. Heat Transfer Engineering, 22(3): 41-51.
Booker, A., Dennis, J., Frank, P., Serafini, D., Torczon, V., & Trosset, M. (1999). A Rigorous
Framework for Optimization of Expensive Functions by Surrogates. Structural
Multidisciplinary Optimization, 17: 1-13. DOI: 10.1007/BF01197708.
Crawley, D., Lawrie, L., Winkelmann, F., Buhl, W., Huang, Y., Pedersen, C., . . . Glazer, J.
(2001). EnergyPlus: Creating a New-Generation Building Energy Simulation Program.
Energy and Buildings, 33: 319-331. DOI: http://dx.doi.org/10.1016/S0378-
7788(00)00114-6.
De Sousa, L., Eykamp, C., Leopold, U., Baume, O., & Braun, C. (2012). iGUESS--A Web Based
System Integrating Urban Energy Planning and Assessment Modelling for Multi-Scale
Spatial Decision Making. 2012 International Confess on Environmental Modelling and
Software. Leipzig, Germany: International Environmental Modelling and Software
Society.
Deru, M., Field, K., Studer, D., Benne, D., Griffith, B., Torcellini, P., . . . Crawley, D. (2011). U.S.
Department of Energy Commercial Reference Building Models of the National Building
Stock. Washington, D.C.: U.S. Department of Energy.
Fumo, N., Mago, P., & Luck, R. (2010). Methodology to Estimate Building Energy Consumption
using EnergyPlus Benchmark Models. Energy and Buildings, 42: 2331-2337. DOI:
http://dx.doi.org/10.1016/j.enbuild.2010.07.027.
Hawkes, A., & Leach, M. (2005). Impacts of Temporal Precision in Optimisation Modelling of
Micro-Combined Heat and Power. Energy, 30(10): 1759-1779. DOI:
http://dx.doi.org/10.1016/j.energy.2004.11.012.
Horlock, J. (1987). Cogeneration, Combined Heat & Power (CHP): Thermodynamics and
Economics. New York: Pergammon Press.
Kavvadias, K., Tosios, A., & Maroulis, Z. (2010). Design of a Combined Heating, Cooling, and
Power System: Sizing, Operation Strategy Selection, and Parametric Analysis. Energy
Conversion and Management, 51: 833-845. DOI: http://dx.doi.org/10.1016/
j.enconman.2009.11.019.
Keirstead, J., Jennings, M., & Sivakumar, A. (2012). A Review of Urban Energy System Models:
Approaches, Challenges, and Opportunities. Renewable and Sustainable Energy
Reviews, 16: 3847-3866. DOI: http://dx.doi.org/10.1016/j.rser.2012.02.047.
Keirstead, J., Samsatli, N., & Shah, N. (2009). SynCity: An Integrated Tool Kit for Urban Energy
Systems Modelling. Fifth Urban Research Symposium 2009. London, England.
Kolanowski, B. F. (2000). Small-Scale Cogeneration Handbook. Lilburn, GA: The Fairmont
Press, Inc.
Best, et al.
If applicable, page number will go here after aggregating all papers
Metso Global. (2014). Metso EtaShred ZZ Zerdirator. Retrieved from Metso:
http://www.metso.com/recycling/mm_recy.nsf/WebWID/WTB-090706-22575-
E7A33?OpenDocument#.U18AD_ldUnF
Organization, U. N. (1998). Seminar on Energy Conservation in Iron Casting Industry. Hanoi,
Vietnam: United Nations Industrial Development Organization.
Payrhuber, K., & Trapp, C. (2011). GE's New Jenbacher Gas Engines with 2-Stage
Turbocharging. Internationale Energiewirtschaftstagung an der TU Wien (pp. 1-14).
Wien, Germany: Technical Universitat Wien.
PHX. (2008). ModelCenter 8.0. Retrieved from Phoenix Integration: http://www.phoenix-int.com/
Pine, G. D. (1979). Assessment of Integrated Urban Energy Options. Cambridge, MA: MIT
Energy Laboratory.
Pohekar, S., & Ramachandran, M. (2004). Application of Multi-Criteria Decision Making to
Sustainable Energy Planning--A Review. Renewable and Sustainable Energy Reviews,
8(4): 365-381. DOI: http://dx.doi.org/10.1016/j.rser.2003.12.007.
Ramsar COP11 Scientific and Technical Review Panel. (2012). Background and Context to the
Development of Principles and Guidance for the Planning and Management of Urban
and Peri-Urban Wetlands. Bucharest, Romania: Ramsar COP11.
Robinson, D., Haldi, F., Kampf, J., Leroux, P., Perez, D., Rasheed, A., & Wilke, U. (2009).
CitySim: Comprehensive Micro-Simulation of Resource Flows for Sustainable Urban
Planning. Building Simulation 2009 (pp. 1083-1090). Glasgow, Scotland: International
Building Performance Simulation Association.
Rue, D. M., Servaites, J., & Wolf, W. (2007). Industrial Glass Bandwidth Analysis. Des Plaines,
IL: Gas Technology Institute.
Sevlian, R., & Rajagopal, R. (2013). Value of Aggregation in Smart Grids. 2013 IEEE
International Conference on Smart Grid Communications (pp. 714-719). Vancouver, BC:
IEEE.
The National Academy of Sciences. (2013). Our Energy System. Retrieved from The National
Academies: http://needtoknow.nas.edu/energy/interactive/our-energy-system/
Therkelsen, P., Masanet, E., & Worrell, E. (2014). Energy Efficiency Opportunities in the U.S.
Commercial Baking Industry. Journal of Food Engineering, 130: 14-22.
Thornton, R. P. (2005). IDEA Report: The District Energy Industry. Westborough, MA:
International District Energy Association.
U.S. Energy Information Administration. (2012). What is the Electric Power Grid, and What are
Some Challenges it Faces? Retrieved from U.S. Energy Information Administration:
http://www.eia.gov/energy_in_brief/article/power_grid.cfm
Van Dam, K., & Keirstead, J. (2010). Re-Use of an Ontology for Modelling Urban Energy
Systems. 2010 Third International Conference on Infrastructure Systems and Services:
Next Generation Infrastrcuture Systems for Eco-Cities (pp. 1-6). Shenzhen, China: IEEE.
Optimizing the Urban Plan of a CCHP-Powered Community for Total Fuel Cycle Efficiency
If applicable, page number will go here after aggregating all papers
Supplementary Information
Optimizing the Total Fuel Cycle Efficiency of an Idealized CCHP-
Powered Community in Oakland, CA
Robert Best Stanford University, robbest@stanford.edu
Forest Flager Stanford University, forest@stanford.edu
Caroline Nowacki Stanford University, cnowacki@stanford.edu
Martin Fischer Stanford University, fischer@stanford.edu
Michael Lepech Stanford University, mlepech@stanford.edu
Appendix 1: Summary of building types simulated for case study including heights in stories, gross floor
area, and annual heating, cooling, and electrical loads in MJ/m
2
.
Building Type Stories
Gross Floor
Area (m
2
)
Heating
(MJ/m
2
)
Cooling
(MJ/m
2
)
Electricity
(MJ/m
2
)
Single Family Residential 1 223 112.7 1.018 214.9
Residential Townhouses 2 392 64.21 71.27 470.5
Mid-Rise Residential Apartments 4 3,135 17.12 119.9 321.7
High-Rise Residential Condominiums 12 9,405 4.864 165.9 285.0
Small Hotel 4 4,013 70.11 51.02 530.7
Large Hotel 6 11,345 79.87 387.7 1,089
Small Office 1 511 12.80 74.98 435.5
Medium Office 3 4,982 50.16 143.5 401.9
Large Office 12 46,320 31.28 119.9 398.7
Warehouse 1 4,835 57.10 1.820 173.1
Refrigerated Warehouse 1 4,835 0.00 580,700 580,700
Bakery 1 4,835 19.67 7.123 647.1
Foundry 1 4,835 1.970 612.0 695.3
Steel Recycler 1 4,835 25.83 5.393 2,972
Glass Factory 1 4,835 41.32 3.642 301.6
Hospital 5 22,422 480.1 796.2 877.4
Outpatient Care 3 3,804 477.8 1,019 905.5
Primary School 1 6,871 34.88 111.1 431.5
Secondary School 2 19,592 19.63 68.66 386.8
Mixed Use: Condos, First Floor Retail 12 9,405 4.35 174.3 287.1
Mixed Use: Offices, First Floor Retail 12 46,320 157.1 593.8 1,961
Strip Mall 1 2,090 133.6 34.15 446.5
Stand-Alone Retail 1 2,294 118.4 31.20 409.2
Quick Service Restaurant 1 232 273.0 81.74 3,201
Full Service Restaurant 1 511 459.0 115.3 2,552
Supermarket 1 4,181 343.0 9.941 1,409
Best, et al.
If applicable, page number will go here after aggregating all papers
Appendix 2: Equipment types and operating parameters simulated for the CCHP plant in this study.
Equipment Type Chosen Model Operating Characteristics
Prime Mover (Heat and Power
Generation Technology)
GE Jenbacher 920 Gas-Fired
Combustion Engine
9.5 MW
0.85 Heat to Power Ratio
7006 Btu/kWh Heat Rate
Max Electrical Efficiency: 0.475
Centrifugal Chiller
McQuay MPV Centrifugal
Compressor Water Chillers
COP: 3.5
Absorption Chiller
York YHN Double-Effect Absorption
Chillers
COP: 1.5
Boiler Standard Industrial Boiler Efficiency: 80%