Grantee Research Project Results
2003 Progress Report: Implications of Climate Change for Regional Air Pollution, Health Effects and Energy Consumption Behavior
EPA Grant Number: R828731Title: Implications of Climate Change for Regional Air Pollution, Health Effects and Energy Consumption Behavior
Investigators: Ellis, Joseph H. , Hobbs, Benjamin F. , Joutz, Frederick L.
Institution: The Johns Hopkins University , George Washington University
Current Institution: The Johns Hopkins University
EPA Project Officer: Chung, Serena
Project Period: September 1, 2000 through August 31, 2003 (Extended to February 7, 2005)
Project Period Covered by this Report: September 1, 2002 through August 31, 2003
Project Amount: $1,376,739
RFA: Assessing the Consequences of Interactions between Human Activities and a Changing Climate (2000) RFA Text | Recipients Lists
Research Category: Climate Change , Air
Objective:
The overall objective
of this research project is to develop a scientifically credible modeling facility
that will help policymakers and analysts understand the effects of human activities
on climate change and variability as well as the possible human responses and
adaptations to climate change and variability. This research program has four
major modeling elements: (1) climate change and variability; (2) electrical
energy demand and production; (3) regional air pollution; and (4) human health
effects associated with air pollution exposure. The overall connections among
the modeling elements listed above are shown schematically in Figure 1.
Figure 1. Overall Connections Among the Modeling Elements
Progress Summary:
Progress is described below and divided into three categories, following the three main modeling activities.
Long-Run Electricity Demand Response to Climate Warming (F. Joutz and C. Crowe, George Washington University)
The short-run statistical analyses summarized in the previous year’s progress report reflected changes in equipment utilization. Long-run shifts in capital stock that could result from consumers shifting to more energy-efficient space conditioning equipment or from greater adoption of air conditioning could not be considered in such analyses. Relative to short-run changes to a given temperature increase, long-run changes could either be larger (if more installations of air conditioning occur) or smaller (if more efficient equipment is substituted for less efficient equipment).
To consider long-run cooling electricity demand and technology choice in response to climate change, it is necessary to use a model that allows capital stock to be a variable. Using the residential, commercial, and industrial electricity demand modules of the National Energy Modeling System (NEMS) (U.S. Energy Information Agency; see Relevant Web Sites), we considered the effects of an increase in summer temperatures of 2°F, 3°F, and 6°F by 2025. These low, midrange, and high scenarios produced increases in the U.S. residential space cooling demand of 10 percent, 14 percent, and 33 percent, respectively. Overall residential energy demand for the three scenarios increased by 1 percent, 2 percent, and 5 percent, respectively. There was no corresponding change in commercial or industrial demand.
Under climate warming, the residential energy mix in 2025 included a higher percentage of renewable sources, particularly geothermal heat pumps. The low, midrange, and high scenarios produced increases in energy derived from geothermal heat pumps of 3 percent, 4 percent, and 10 percent. Again, there was no corresponding change for the commercial or industrial sectors. This is partly because these latter modules assume a 30 percent discount rate for new appliance technology adoption, reflecting the historically high payback requirements for energy efficiency investments. Education and “market transformation” initiatives such as the U.S. Environmental Protection Agency’s (EPA) “Energy Star” program, however, might shift the required paybacks downward. Reducing this discount rate may produce an increase in renewable energy demand outside the residential module. We are investigating several other specification parameters, and we are seeking to identify elements of the residential and commercial energy consumption models that are “hard wired” into the NEMS or are otherwise not free to adjust with the forecast.
We are developing disaggregated responses by the North American Electric Reliability
Council (NERC)/electric region. This requires an additional procedure beyond
the NEMS forecasts because the U.S. census regions used in the NEMS differ somewhat
from the NERC/electric regions. It also will be necessary to disaggregate the
demand responses by load demand period for use in the electricity market simulation
models.
Short-Run Electricity Market Response to Changed Temperatures and Demands (B.F.
Hobbs and Y. Chen, Johns Hopkins University)
The objective of this analysis is to update the analysis of temperature sensitivity on electric sector pollution emission based on the load temperature sensitivities provided by the research group from George Washington University and assumed changes in electric generator efficiency as a result of temperature increases. The focus is on the short-run (fixed-generation installations) response of the Pennsylvania-Jersey-Maryland (PJM) system.
Procedures of Analysis. The assumed sensitivities of generator characteristics (megawatts of electrical output [MWe] capacity and MBTU/kWh heat rate) to temperature changes are the same as in previous progress reports. We perform episode analysis by the following steps:
· Including the regional NOx trading cap in our model, our analysis focuses on the ozone season of May 1 to September 30, 2000, a total of 3,672 hours. Excluding for the 3-day ozone episode period considered in the second step, the load distribution of the remaining 3,600 hours is approximated by five load periods. The PJM market operation is determined separately within each of those periods in a later step. The peak period has 60 hours, whereas the other periods have 885 hours each. Together with the 72 periods (1 for each hour) for the episode days, our model includes 77 periods. These loads are assigned to each of 14 nodes in an aggregated PJM network.
· We assessed candidates for ozone episodes by looking at 3-day moving averages of temperature in Philadelphia and Washington, DC, during July and August of 2000. With a moving average of 82°F, we selected August 7 to August 9, 2003, a span of 72 hours, as our episode period.
· The year 2000 loads obtained from the PJM Web Site serve as the base case loads. To construct a load scenario under climate warming, we increase the base case load by hourly-and-node specific factors calculated by the George Washington University group (see previous progress reports). The resulting loads represent our climate-sensitive load.
· We then apply a least-cost production costing linear programming model,
constructed in Years 1 and 2 of the project, to simulate the PJM market response
to the assumed loads. This assumes that the market is reasonably competitive,
which the PJM Market Monitoring Unit confirms was the case for that year. As
a check, we also simulated oligopolistic (Cournot) outcomes for that period
and found that the actual prices in 2000 were most consistent with competitive
conditions, or equivalently were closest to prices assuming that generators
in PJM were heavily forward-contracted, which dampens incentives to restrict
output and raise prices in the forward market. This was indeed the case with
PJM generators in 2000, which accounts for the relative competitiveness of that
market relative to California in that year. Chen and Hobbs (2004) describe the
competitive and oligopoly simulations of PJM under the NOx cap for the year
2000 ozone season. The outputs of the market simulation model include generator
output by source and time period (MW) and each generator’s NOx and SO2
emissions by time period (tons).
Summary of Results. We summarize the impact of a 2°F increase in
ambient temperature on pollution emission for ozone season in the next sections.
In particular, we look at the overall impact on the entire ozone season as well
as the hourly emission change during the hypothesized 3-day episode.
The updated load sensitivity shows that the overall impact on ozone season is an average 4.3 percent increase of load over the 2000 ozone season as a result of a 2°F increase in ambient temperature. This increase of demand leads to no change of total NOx emissions—by definition—because emissions are capped. Substantial impacts on fuel cost, however, are observed among scenarios. The increase in fuel cost compared with the base scenario is 21 percent, 0.4 percent, and 22 percent as a result of load increase alone, deterioration in generator efficiency caused by temperature, and both effects together, respectively, for the entire ozone season.
Turning to the 3-day assumed ozone episode, Figure 2 shows the sum over nodes of loads for the 3-day episode period. Table 1 summarizes the emissions and fuel cost impacts. The impacts caused by load changes are orders of magnitude greater than the impacts caused by changes in generator efficiency. Note that for this 3-day period, NOx emission impacts are not zero because emissions are capped only for the entire ozone season and not for particular days. NOx emissions have increased during this time, implying that emissions at other times are lower to meet the overall cap.
Figure 2. Load Profile Over 3-Day Episode Period (August 7-9)
Table 1. Electricity Demand and Generators’ Performance
Impact Relative to Base Case for 3-Day Ozone Episode, Assuming 2oF Temperature
Increase
Impact/Category | NOx Emission | SO2 Emission | Fuel Cost |
Demand Impact | 5 % | 5.5 % | 19.9 % |
Generators’ Performance Impact | 0.076 % | -0.001 % | 0.25 % |
Joint Impact | 4.9 % | 5.4 % | 20.3 % |
Figure 3 illustrates the hourly emissions impact during our 3-day episode period;
the left graph represents the NOx emission and the right one is for
SO2 emission. The climate change scenario is the joint impact of
load increase and generators’ efficiency deterioration caused by the ambient
temperature increase. The average impact is 4.9 percent and 5.4 percent for
NOx and SO2, respectively. Although most NOx
emission increases occur during peakload hours (i.e., daytime), the increase
of SO2 emission tends to occur during both daytime and nighttime.
Figure 3. NOx and SO2 Emissions Impact During 3-Day Episode Period, PJM
Figure 4 further decomposes the aggregate emissions impact of the 3-day episode into state-level results. In terms of NOx emissions, the greatest tonnage impact occurs in Maryland and Pennsylvania, amounting to 6.3 percent and 3.3 percent increases compared with base case, respectively. The percentage increases for NOx in Delaware and New Jersey is 10.1 percent and 3.5 percent, respectively, but the tonnages involved are much smaller than the other states. Among states, the SO2 tonnage impact is highest in Pennsylvania and lowest in Maryland.
Figure 4. State-Level Impact of 2°F Warming During 3-Day Episode
Analysis of Renewable Portfolio, Emissions Cap, and Green Pricing Tradeoffs
There is a range of policies in place to limit the environmental impacts of the electric power sector, which might act to mitigate the effects of climate warming on the sector’s emissions. An analysis of their interacting effects was conducted using the PJM oligopoly model to understand better how the policies affect costs and emissions, particularly if generators engage in strategic behavior in the various markets.
Renewable portfolio standards and green pricing programs are two distinct approaches in power markets for promoting renewable generation. The cost of emissions allowances also can be an incentive to install renewables. The renewable portfolio standard is a mandatory requirement that a fixed percent of power delivered to customers from suppliers (or load service entities) has to be from renewable sources. In contrast, green pricing programs offer a voluntary opportunity for customers with a higher willingness-to-pay associated with environmental good to pay a premium to procure their power from renewable sources. An emerging issue is the interaction of green pricing programs, renewable portfolio standards, and allowances markets. Such interactions possibly may decrease the competiveness of power markets and induce inefficiencies and welfare loss. An oligopolistic equilibrium model based on a complementarity formulation was applied to investigate such interactions. In the model, renewable generation is modeled as a differentiated product for which consumers have a demand curve, with assumed cross-elasticities relative to the demand for so-called “grey” (nonrenewable) energy. The renewable portfolio standards are formulated as a coupled constraint imposed over the compliance period with tradable credits. Suppliers with a substantial capacity share are designated as strategic players in the respective markets, exercising a Cournot strategy, and the remaining capacity is treated as a competitive fringe.
The application was to the EPA NOx Budget Program and PJM market during the 2000 ozone season, using the same input data as was used in the climate impact analysis. The results show that total energy consumption is constrained by renewable generation caused by renewable portfolio standards. Substantial market power can be introduced if a supplier simultaneously exercises market power in the renewable and conventional power markets. In comparison to a scenario in which suppliers only possess market power in the conventional power market, the power prices are higher, given that Cournot suppliers concurrently exercise market power in both power markets. The NOx allowance price ,however, is lower because the market power in the renewable market suppresses the demand for allowances.
Regional Air Pollution Modeling and Health Effects Characterization–J.H. Ellis
There have been dramatic changes in the system with which we generate ambient air concentration fields for use in this project. This has had positive and negative effects, with the former far outweighing the latter. First the negative effect: making the new system operational has been very time- and labor-intensive and has slowed this year’s progress in this portion of the project considerably. The benefits associated with this development are, however, very large and wide ranging.
Prior to this year, we were restricted (both hardware- and software-driven) to performing limited duration scenarios (on the order of several days long) and limited (spatial) domain scenarios. We now can successfully generate continental U.S. scenarios for any arbitrary run length. For example, several months ago we completed a suite of scenarios using the Mesoscale Model 5 (MM5)/Meteorology-Chemistry Interface Processor (MCIP)/Sparse Matrix Operator Kernel Emissions (SMOKE) Model/Community Multiscale Air Quality (CMAQ) Model for the entire United States. The scenarios also were performed for the period of May 1 through September 30 for 1990-1999 and 2050-2059 (this latter group of 10 years of ozone season simulations used Goddard Institute for Space Studies (GISS) output as input to regridder in the MM5). In the MM5, these runs first were made for a 108 km domain and then nested to 36 km, which subsequently was fed to the MCIP, SMOKE, and CMAQ models.
Most recently, we reran the MCIP analyses to produce MCIP output files sufficiently small (i.e., daily) that NCO data extraction utilities (e.g., ncks) would function. The issue here was extraction of surface temperature data required in the energy demand and distribution analyses described above.
Specifics regarding the new hardware system follow:
· Master node (dual Xeon CPU), 3 Gb memory, 800 Gb U320 SCSI internal disk capacity, dual gigabit ethernet, 200/400 Gb LTO-2 tape backup).
· Forty compute nodes (each with dual Xeon CPU, 1 Gb memory, 150 Gb U320 SCSI internal disk capacity, gigabit ethernet).
· Four network-attached storage devices (each with single P4 CPU, 1 Gb memory, 1,000 Gb fast IDE disk capacity, gigabit ethernet).
· Master node-attached 2,100 Gb U320 SCSI disk array.
· Twenty-four port-managed gigabit ethernet switch (private, secure network).
· Two 16-port unmanaged gigabit ethernet switches.
· Two 32-port keyboard-video-mouse (KVM) switches.
· Two 42U rack mount enclosures.
· Seven 3,000VA UPS.
· Cat5 KVM extenders (allowing multiple displays to control the cluster).
· Portland Group f77, f90, and cc compilers, NCAR Graphics, PAVE Visualization System, MPI.
· Linux RedHat9.0 operating system.
We recently applied for and were awarded $20,000 by the Engineering School to create a computer room dedicated to this machine (seven 30A circuits, 60,000 BTU/hour cooling capacity–the cluster is noisy and it is hot). This renovation work will commence mid-August 2004.
On the software side, all of the SMOKE and CMAQ analyses now are performed 1 day at a time (the MM5 is still run as a single large job—a 5-month, U.S.-wide run produces an MM5 output file size of about 70 Gb). I have benefited greatly from the generous assistance of Christian Hogrefe in making this software operational (it was he who originally provided us the various daily scripts that we subsequently modified for this work). In addition, our colleagues at Columbia University and the National Aeronautics and Space Administration have been extremely helpful in providing and making operational the GISS output used for future-year runs.
Journal Articles on this Report : 3 Displayed | Download in RIS Format
Other project views: | All 24 publications | 10 publications in selected types | All 9 journal articles |
---|
Type | Citation | ||
---|---|---|---|
|
Bell ML, Hobbs BF, Ellis H. Metrics matter: conflicting air quality rankings from different indices of air pollution. Journal of the Air & Waste Management Association 2005;55(1):97-106. |
R828731 (2003) R828731 (Final) R828733 (Final) |
Exit Exit |
|
Chen Y, Hobbs BF, Leyffer S, Munson TS. Leader-follower equilibria for electric power and NOx allowances markets. Computational Management Science 2006;3(4):307-330. |
R828731 (2003) R828731 (Final) R831836 (2005) R831836 (2007) R831836 (2008) R831836 (Final) |
Exit |
|
Crowley C, Joutz FL. Hourly electricity loads: temperature elasticities and climate change. Energy Studies Review. |
R828731 (2003) |
not available |
Supplemental Keywords:
air toxics, climate change, particulate matter, tropospheric ozone, ambient air pollution, climate models, climate variability, climate variations, ecosystem sustainability, electrical energy, emissions inventory, energy generation, exposure and effects, human activities, human activity, human exposure, integrated assessments, policymaking., RFA, Scientific Discipline, Health, Air, Geographic Area, particulate matter, air toxics, Health Risk Assessment, climate change, State, Risk Assessments, Atmospheric Sciences, tropospheric ozone, integrated assessments, electrical energy, PM10, environmental monitoring, exposure and effects, stratospheric ozone, policy making, Virginia (VA), Delaware (DE), human activities, PM 2.5, energy generation, climate variations, climate models, Maryland (MD), emissions inventory, human exposure, DC, PM, ecosystem sustainability, human activity, climate variability, ambient air pollution, Global Climate ChangeRelevant Websites:
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.