Dynamic Geospatial Modeling of the Building Stock To Project Urban

Jun 26, 2018 - ABSTRACT: In the United States, buildings account for more than 40% of total energy consumption and the evolution of the urban form wil...
0 downloads 0 Views 4MB Size
Download PDF
Policy Analysis Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX

pubs.acs.org/est

Dynamic Geospatial Modeling of the Building Stock To Project Urban Energy Demand Hanna M. Breunig,*,† Tyler Huntington,‡ Ling Jin,† Alastair Robinson,† and Corinne D. Scown†,‡ †

Energy Technologies Area, Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, California 94720, United States Joint BioEnergy Institute, Emeryville, California 94608, United States

‡

Downloaded via NEW MEXICO STATE UNIV on July 3, 2018 at 19:35:27 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: In the United States, buildings account for more than 40% of total energy consumption and the evolution of the urban form will impact the effectiveness of strategies to reduce energy use and mitigate emissions. This paper presents a broadly applicable approach for modeling future commercial, residential, and industrial floorspace, thermal consumption (heating and cooling), and associated GHG emissions at the tax assessor land parcel level. The approach accounts for changing building standards and retrofitting, climate change, and trends in housing and industry. We demonstrate the automated workflow for California and project building stock, thermal energy consumption, and associated GHG emissions out to 2050. Our results suggest that if buildings in California have long lifespans, and minimal energy efficiency improvements compared to building codes reflective of 2008, then the state will face a 20% or higher increase in thermal energy consumption by 2050. Baseline annual GHG emissions associated with thermal energy consumption in the modeled building stock in 2016 is 34% below 1990 levels (110 Mt CO2eq/y). While the 2020 targets for the reduction of GHG emissions set by the California Senate Bill 350 have already been met, none of our scenarios achieve >80% reduction from 1990 levels by 2050, despite assuming an 86% reduction in electricity carbon intensity in our “Low Carbon” scenario. The results highlight the challenge California faces in meeting its new energy efficiency targets unless the State’s building stock undergoes timely and strategic turnover, paired with deep retrofitting of existing buildings and natural gas equipment.

■

INTRODUCTION With buildings accounting for more than 40% of total U.S. energy consumption,1 understanding their collective changes across time and space is critical to predicting and managing future energy demand. Evaluating the potential for energy savings from technologies such as District Energy Systems (DES), zero net energy (ZNE) homes, smart windows, and other building and equipment retrofits requires data on current and expected future spatial distributions of energy demand as well as the locations of anticipated new construction. Spatially resolved building stock projections can also provide useful data for studying urban climate and air quality phenomena, including the urban heat island effect, but frameworks for developing these projections are lacking.2−6 While no model will capture all the forces that influence urban energy demand, developing a robust estimation framework that is amenable to sensitivity analyses can lead to valuable insights that inform research as well as stateand local-level decision-making. Gross potential for deployment of energy efficient technologies in urban settings can be estimated from total floorspace of target building types in a study area,7 but this simplified approach can over- or underestimate potential. Additional knowledge of building turnover at fine spatial resolutions is © XXXX American Chemical Society

needed to match energy-saving strategies for buildings with the communities where they will be most environmentally and economically beneficial.8−10 Digitized maps of land parcels and associated tax-assessor building information have been used to represent the existing urban form in many areas of the United States (U.S.). Parcel data was used by Heiple and Sailor in their analysis of building energy consumption for Houston, Texas, and by Pincetl et al. in their analysis of building energy consumption and CO2 emissions in Los Angeles, California.2,8 A number of approaches exist for estimating the energy intensity of building stocks.11,12 Bottom-up approaches convert a floorstock into energy consumption using prototype building energy use intensities (EUI) normalized by floorspace [kWh/ft2 or kW/ ft2].3 However, identifying published EUIs representative of the buildings in a study area (typically produced using utility customer data or individual building simulation) can be a challenge and many city-level studies ultimately rely on topdown approaches for estimating energy consumption. Such Received: Revised: Accepted: Published: A

January 23, 2018 June 5, 2018 June 26, 2018 June 26, 2018 DOI: 10.1021/acs.est.8b00435 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Policy Analysis

Environmental Science & Technology

(MF) residential floorspace can be determined from changes in population, densification (fraction of residential construction that is MF versus single family), and household size (occupants); (3) buildings are demolished when they reach the end of their expected average lifespan (specific to each use type); (4) a building built on a parcel where a building was demolished (a “rebuild”) will have the same use type as the previous building and will be the same size or larger; (5) if employment decreases in a county, buildings are not demolished but left vacant or repurposed. Land and building attributes collected by California Tax Assessor County Offices were acquired as geospatial files from the data vendor ParcelQuest for the input year 2016 (T0). For this study, parcels are classified into 19 representative commercial, industrial, and residential building use types and 10 building vintages, from which floorspace totals are derived, using the following building attributes: use code; build year; effective build year (indicating a major retrofit occurred) (Supporting Information Section 1 and Table S1).18 Drivers of Building Stock Dynamics. Changes in population growth drives floorspace demand at regional scales.19,20 As employment data has become available at finer industrial sector and spatial scales, a body of literature has emerged that uses employment to estimate demand for new office space.21 Change in employment is not a perfect proxy for change in active floorspace, as rent is relatively cheap compared to labor costs (i.e., a company that expands or downsizes its staff may be slow to move to a more suitably sized building) (key assumption 5). However, it is currently the best indicator given the improved quality of employment data at the sector level, and the strong evidence suggesting employment eventually drives active floorspace growth.21 This analysis relies on county-level working population (i.e., employment) as a predictor of active commercial and industrial floorspace (key assumption 1), whereby this floorspace is met by prioritizing the replacement of buildings scheduled for demolition with larger buildings of the same use type, and the activation of vacant land and buildings (details in the Supporting Information Section 2). Prioritizing rebuilds (key assumption 4) reflects our assumption that existing buildings are built on higher-value land and are appealing for new construction projects and follows the historical trend in the U.S. of buildings getting larger.19 The Employment Development Department routinely publishes employment projections for industries at the county level. Industries in these projections are matched with building use types modeled in this analysis (Table S4), and a linear regression is performed to estimate growth from the most recent projection year (2024) out to T2 (2050). The growth in active floorspace for each building type and county is assumed to match growth in the associated employment industries, with changes capped at ±15% for 2020 or +200% and −50% for 2050. These limits reflect feasible growth based on historical national floorspace expansion (e.g., even with a recession, commercial floorspace increased by 21% between 2003 and 2012),19 and the allocation of floorspace to vacant buildings (i.e., not all increases in active floorspace will be met by new construction). Projected growth is reduced by 50% in Sensitivity Run 3 to reflect a weaker association between employment and active floorspace growth. Change in MF floorspace is estimated at the county-level using an approach adapted from McCarthy et al., which projects growth in the California single-family (SF) and MF housing stock.20,22 Following their method, total household projections are adapted from the California Department of Finance using

approaches use energy consumption data aggregated by utilities or government surveys at coarse spatial scales to interpolate demand at higher spatiotemporal resolution.13 For example, percapita energy consumption has been resolved to finer spatial and temporal scales using diurnal population maps.14 By focusing on static approximations, both bottom-up and top-down approaches have limited usefulness in long-term planning unless they are integrated with a building stock turnover model and EUIs representative of future buildings. The low spatial resolution of previous energy demand forecasts makes it challenging for policy makers and utility resource planners to predict the location and magnitude of future infrastructure constraints or excess capacity. In this paper, we present a novel methodology for modeling parcel-level dynamics in building stocks. We demonstrate this strategy by projecting California’s building stock to the nearterm (2020) and midterm (2050) and analyze the predicted evolution of thermal consumption (heating and cooling) and associated greenhouse gas (GHG) emissions in commercial, residential, and industrial buildings. Our approach combines top-down strategies for estimating future floorspace with bottom-up strategies for estimating current and future energy use in a workflow automated in R.15 We use California as a case study to demonstrate how this methodology can be used to understand key drivers of local energy consumption and to identify opportunities for deploying energy efficient measures. This model can be applied to other states where building and energy data is available. A table listing data requirements and resources for the U.S. and other countries is included in the Supporting Information. In 2017, the California Energy Commission (CEC) announced new targets for doubling energy efficiency savings by 2030, as mandated by Senate Bill 350, to support GHG emissions reductions of 40% below 1990 levels by 2030 and 80% below by 2050.16 These targets are heavily reliant on widespread retrofitting of existing buildings and replacement of old appliances. Additionally, the Title 24 Building Energy Efficiency Standards that go in effect in 2020 requires newly constructed residential buildings of three floors or fewer to install solar photovoltaic systems (PVS). As of 2015, approximately 98% of single family homes in the Pacific region have fewer than three floors and 26% of apartments have less than five units and may fall within the new standard’s height cutoff.17 While our model could be modified to capture the effects of PVS adoption in these segments, the analysis presented in this paper focuses on the estimation of GHG emissions from multifamily, commercial, and industrial buildings that are likely to be exempt.

■

MATERIALS AND METHODS Building Stock Dynamics Model. Building stock dynamics typically refers to the processes of construction, demolition, and retrofitting of floorspace over time but can also include changes in use type or vacancy. In this analysis, changes in floorspace are first estimated at the county level using aggregated data from a base year (T0), building stock, along with multipliers discussed in the following sections, and then allocated to parcels through a hierarchical process that prioritizes replacing old buildings and vacant lots over new construction on undeveloped land. This approach sequentially models each projection year so that changes to parcels in 1 year are reflected in future years. The model relies on several key simplifications: (1) active commercial and industrial floorspace changes proportionally with sector-specific employment rates; (2) active multifamily B

DOI: 10.1021/acs.est.8b00435 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Policy Analysis

Environmental Science & Technology

Effective build years are adjusted to T1. Multipliers less than 1 indicate some economic growth but not enough growth to warrant rebuilds of all demolished buildings. For counties and building types with multipliers less than 1, the positive additional floorspace is met by selecting parcels with the largest demolished floorspace and multiplying the demolished floorspace by a factor between 1.2 and 4, increasing by 0.2, until the required county level floorspace addition is met. If not met, the model moves on to the parcel with the next largest demolished floorspace and assigns rebuilt floorspace in the same manner. This assumes that land with large demolished buildings is more likely to be rebuilt than land with small buildings.

the department’s county population projections and personsper-household statistics (key assumption 2).22,23 Net additional households are split into SF and MF new construction according to trends published by the annual publications of California Construction Review. Data from the U.S. Census Bureau American Community Survey (ACS) and the EIA Residential Energy Consumption Survey (RECS) database could be used to approximate MF floorspace in other locations.24,25 Assuming a specific floorspace decay rate, as done in regional or national analysis, does not translate to parcel-level building stock changes. While data is limited and can vary due to owner decisions and climate, surveys of U.S. building stock suggest that buildings built for a specific use type have an average lifespan before they are demolished (key assumption 3).19,26,27 As a starting point, we assume buildings in California survive 40 years longer than the average national buildings of the same type due to milder climate conditions and the general age of buildings in the 2016 parcel data (Table S1).26,28 We change the average lifespan to the national averages in Sensitivity Run 1. Parcel-Level Floorspace Allocation. Floorspace reported in the T0 parcels is aggregated by building type and county to represent the T0 building stock. Values are multiplied with the county-level growth rates for each building use type to determine total floorspace in T1 (eq 1): ij yz T1 Floorspacei , j = jjjj∑ T0 Floorspacei , j , k zzzz × ri , j j k z k {

RebuildMultiplieri , j =

(∑ T DemolishedFloorspace ) k

0

i ,j,k

if Multiplier is ≤ 4 and ≥ 1

(3)

In cases where additional floorspace is required to meet expected county-level growth, floorspace is allocated to parcels with empty lot areas and vacant buildings, in that order. This prioritization leverages the fact that empty lots are assigned a residential, commercial, or industrial use code, giving some insight as to the suitability of the land for a specific building type, while “vacant” buildings have no other information provided. Lot areas are first converted to allowable building sizes using the median floorspace-to-lot area ratio (BtoLi,j) for that building type and county (eq 4). Then, allowable floorspace in empty lots is allocated to meet the positive additional floorspace required, and build years are adjusted to T1. Empty lots that have some building information (e.g., garage) are prioritized to help capture existing buildings that are missing floorspace information or empty lots that previously included a building and are likely located on higher-value land. Following the exhaustion of lot areas, parcels with vacant use types and with remaining floorspace in T1 are selected to meet positive additional floorspace, with preference given to newer and larger buildings. Then, floorspace is allocated to parcels with vacant use type and no reported building floorspace. Any remaining floorspace is recorded as additional greenfield “development” but not allocated to parcels. It is assumed that nonindustrial buildings will not be converted to industrial use types, so additional floorspace needs for industrial buildings are only met with rebuilds, construction on industrial open lots, and greenfield development. The remaining floorspace that remains unmapped varied from 80% reduction from 1990 levels by 2050, despite assuming an 86% reduction in carbon intensity of the grid in our “Low Carbon” scenario. The replacement of natural gas boilers for heating with technologies that generate lower GHG emissions, in addition to substantial building turnover and retrofitting is key to achieving the state’s 2050 target. Future work is needed to improve the representation of differences between urban/rural drivers by collecting and integrating data not presently captured in government employment databases. Drivers may include analysis of transportationoriented development in key corridors (e.g., high speed rail construction boosting development in rural areas) and autonomous vehicle deployment. Important extensions of this research may also include explicit modeling of equipment electrification and market penetration, electric vehicle adoption,

Figure 5. Maps of county level variations from base case due to sensitivity model runs. Ratio of model variation to base case shown as percentages. Sensitivities of cooling (C) and heating (H) energy consumption are displayed separately under each scenario. Floorspace is abbreviated “flsp”.

have moderate expected improvements in heating efficiencies, such as residential buildings, can lead to county-level increases in heating consumption. Adjustments used to simulate densification (changing rebuild cap from 4 to 8) did not substantially affect state-level results, while a sprawl simulation (changing rebuild cap from 4 to 2) increased thermal consumption by 7%. Factors leading to this increase included floorspace moving into different climate zones, and the creation of active office floorspace in larger buildings, which have noticeably higher EUIs than small offices (Figure 1), reflecting differences such as annual occupancy hours, density of electrical loads and occupant density. As seen in Figure 4, the simulation of densification and sprawl look similar to the base run in dense areas like downtown San Francisco, where floorspace would always be targeted for rebuilds.

■

DISCUSSION Urban development is driven by a confluence of socio-economic and geophysical factors that generates patterns in floorspace.37 The social and environmental consequences of this urban sprawl, such as car-dependency, have been extensively researched.38 In this paper, we present and demonstrate a novel methodology for modeling and mapping dynamics of building stock turnover, thermal energy consumption, and H

DOI: 10.1021/acs.est.8b00435 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Policy Analysis

Environmental Science & Technology

(11) Swan, L. G.; Ugursal, V. I. Modeling of end-use energy consumption in the residential sector: A review of modeling techniques. Renewable Sustainable Energy Rev. 2009, 13 (8), 1819−1835. (12) Kavgic, M.; Mavrogianni, A.; Mumovic, D.; Summerfield, A.; Stevanovic, Z.; Djurovic-Petrovic, M. A review of bottom-up building stock models for energy consumption in the residential sector. Building and environment 2010, 45 (7), 1683−1697. (13) Gils, H. C.; Cofala, J.; Wagner, F.; Schöpp, W. GIS-based assessment of the district heating potential in the USA. Energy 2013, 58, 318−329. (14) Sailor, D. J.; Lu, L. A top−down methodology for developing diurnal and seasonal anthropogenic heating profiles for urban areas. Atmos. Environ. 2004, 38 (17), 2737−2748. (15) Team, R. C. R: A Language and Environment for Statistical Computing; The R Foundation for Statistical Computing: Vienna, Austria, 2016. (16) California Energy Commission. Energy Commission Adopts Targets for Energy Efficiency Savings, 2017. (17) U.S. Energy Information Administration. 2015 Residential Energy Consumption Survey: Energy Consumption and Expenditures Tables, Table CE2.1 Household Site Fuel Consumption in the U.S., 2015. (18) Geospatial and non-geospatial non-property owner attributes for land parcels from all 58 California county tax-assessor offices, ParcelQuest: Folsom, CA, 2016. (19) U.S. Energy Information Agency. Commercial buildings Energy Consumption Survey (CBECS), Manufacturing Energy Consumption Survey (MECS); Residential Energy Consumption Survey (RECS), 2017. (20) McCarthy, R.; Yang, C.; Ogden, J. M. California Energy Demand Scenario Projections to 2050, Institute of Transportation Studies, 2008. (21) Miller, N. G. Workplace trends in office space. Journal of Corporate Real Estate 2012, 16, 159−181, DOI: 10.1108/JCRE-072013-0016. (22) California Energy Commission. Energy Demand Forecast Methods Report, 2005. (23) California Department of Finance. P-1: State Population Projections (2010−2060); P-4: State and County Projected Households, Household Population and Persons per Household, 2015−2016. (24) U.S. Energy Information Administration. Residential Energy Consumption Survey. (25) U.S. Census Bureau. American Community Survey. (26) U.S. Energy Information Agency. Assumptions to the Annual Energy Outlook 2016, 2017. (27) Reyna, J. L.; Chester, M. V. Energy efficiency to reduce residential electricity and natural gas use under climate change. Nat. Commun. 2017, 8, 14916. (28) Reyna, J. L.; Chester, M. V. The Growth of Urban Building Stock: Unintended Lock-in and Embedded Environmental Effects. J. Ind. Ecol. 2015, 19 (4), 524−537. (29) Kavalec, C.; Fugate, N.; Garcia, C.; Soltani Nia, M. California Energy Demand 2016−2026, Revised Electricity Forecast, Vol. 1: Statewide Electricity Demand and Energy Efficiency; California Energy Commission: Sacramento, CA, 2016. (30) DOE Advanced Manufacturing Office. Manufacturing Energy and Carbon Footprints. (31) Nelson, J.; Mileva, A.; Johnston, J.; Kammen, D. M.; Wei, M.; Greenblatt, J. Scenarios for deep carbon emission reductions from electricity by 2050 in western North America using the SWITCH electric power sector planning model: California’s Carbon Challenge Phase II Vol. II; Lawrence Berkeley National Laboratory, 2014. (32) California Air Resources Board. California Greenhouse Gas Emissions Inventory, 2017 ed.; 2017. (33) Environmental Protection Agency. Compilation of air pollutant emissions factors, Vol. 1: stationary point and area sources; 1995; p 10. (34) California Commercial End-use Survey (CEUS). Itron, 2006. (35) Schwartz, L.; Wei, M.; Morrow, W.; Deason, J.; Schiller, S.; Leventis, G.; Smith, S.; Ling Leow, W.; Levin, T.; Plotkin, S.; Zhoe, Y.; Teng, J. Electricity end use, energy efficiency, and distributed energy resources baseline; 2017.

as well as dynamic geospatial modeling of building stock water demand and water heating and cooling energy savings opportunities.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.8b00435. Additional details on data processing and collection, assumptions, and methodologies (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Phone: (510) 486-4046. E-mail: [email protected]. ORCID

Hanna M. Breunig: 0000-0002-4727-424X Corinne D. Scown: 0000-0003-2078-1126 Notes

The authors declare no competing financial interest. Results and aggregate data that supports the plots within this paper and the program developed in R which automates the workflow are available from the corresponding author upon reasonable request.

■

ACKNOWLEDGMENTS This work was supported by the California Energy Commission. This manuscript has been authored by an author at Lawrence Berkeley National Laboratory under Contract No. DE-AC0205CH11231 with the U.S. Department of Energy.

■

REFERENCES

(1) How much energy is consumed in the U.S. residential and commercial buildings?https://www.eia.gov/tools/faqs/faq.php?id=86&t=1 (accessed October 19, 2017). (2) Heiple, S.; Sailor, D. J. Using building energy simulation and geospatial modeling techniques to determine high resolution building sector energy consumption profiles. Energy and Buildings 2008, 40 (8), 1426−1436. (3) Huang, Y. J.; Brodrick, J. A bottom-up engineering estimate of the aggregate heating and cooling loads of the entire US building stock. In Building Industry Trends 10; Lawrence Berkeley National Laboratory, 2000; pp 135−148. (4) Yamaguchi, Y.; Shimoda, Y.; Mizuno, M. Proposal of a modeling approach considering urban form for evaluation of city level energy management. Energy and Buildings 2007, 39 (5), 580−592. (5) Hu, K.; Yang, X.; Zhong, J.; Fei, F.; Qi, J. Spatially explicit mapping of heat health risk utilizing environmental and socioeconomic data. Environ. Sci. Technol. 2017, 51 (3), 1498−1507. (6) Howard, B.; Parshall, L.; Thompson, J.; Hammer, S.; Dickinson, J.; Modi, V. Spatial distribution of urban building energy consumption by end use. Energy and Buildings 2012, 45, 141−151. (7) Long, J.; John, M.; Greenblatt, J.; Wei, M.; Yang, C.; Richter, B.; Hannegan, B.; Youngs, H. California’s Energy Future: The View to 2050; California Council on Science and Technology, 2011; p 66. (8) Pincetl, S.; Graham, R.; Murphy, S.; Sivaraman, D. Analysis of High-Resolution Utility Data for Understanding Energy Use in Urban Systems: The Case of Los Angeles, California. J. Ind. Ecol. 2016, 20 (1), 166−178. (9) Coffey, B.; Borgeson, S.; Selkowitz, S.; Apte, J.; Mathew, P.; Haves, P. Towards a very low-energy building stock: modelling the US commercial building sector to support policy and innovation planning. Building Research & Information 2009, 37 (5−6), 610−624. (10) Asensio, O. I.; Delmas, M. A. The effectiveness of US energy efficiency building labels. Nature Energy 2017, 2 (4), 17033. I

DOI: 10.1021/acs.est.8b00435 Environ. Sci. Technol. XXXX, XXX, XXX−XXX

Policy Analysis

Environmental Science & Technology (36) King, M. Community Energy: Planning, Development and Delivery; 2012. (37) Squires, G. D. Urban sprawl: Causes, consequences, & policy responses. The Urban Insitute, 2002. (38) Johnson, M. P. Environmental impacts of urban sprawl: a survey of the literature and proposed research agenda. Environment and planning A 2001, 33 (4), 717−735. (39) Office of Policy Development and Research. The feasibility of developing a national parcel database: County Data Records Project Final Report, U.S. Department of Housing and Urban Development, ABT Associates Inc, Fairview Industries: Washington, DC, 2013.

J

DOI: 10.1021/acs.est.8b00435 Environ. Sci. Technol. XXXX, XXX, XXX−XXX