5th Annual CMAS Conference Friday Center Chapel Hill, NC 10/17/2006 Process Analysis Application: The 2000 Houston Ozone Attainment Simulations Marianthi-Anna Kioumourtzoglou, William Vizuete, Harvey Jeffries, Barron Henderson vizuete@unc.edu 5th Annual CMAS Conference Friday Center Chapel Hill, NC 10/17/2006 Hello, my name is Marianthi and I will present a work done by William Vizuete, Harvey Jeffries, Barron Henderson and me, on a Process Analysis Application on the 2000 Houston Ozone Attainment simulations. The presentation can be found in this site. http://ftpozone.sph.unc.edu

Outline Introduction Ozone chemistry The need for a tool to quantify physical and chemical processes for better understanding of O3 formation (Process Analysis in Python) Application Here is the outline of my presentation. In the beginning there will be a small introduction to the subject, followed by a simplified overview of the ozone chemistry. This will lead to the need for a tool to quantify physical and chemical processes for better understanding of the O3 formation (the tool is called pyPA, Process Analysis in Python). In the end I am going to present an application for the pyPA tool.

Introduction Region of Interest: Houston, TX, designated as Non-Attainment Region by EPA Model used: Comprehensive Air Quality Model with extensions (CAMx) The Emission Inventory used was developed by TCEQ for the 8hr SIP modeling scenario Post Processor: Process Analysis in Python Houston is the 4th largest city in the US. The Environmental Protection Agency (EPA) has designated the Houston region as NON-ATTAINMENT area for the 1-hr ozone standard as severe and for the recently implemented 8-hr standard (as moderate). Ozone is a secondary pollutant with very complicated non-linear chemistry. That is why we use photochemical models to understand this chemistry. The model used for this analysis was the Comprehensive Air Quality Model with extensions (CAMx) and the emission inventory used was the one developed by the Texas Commission on Environmental Quality for the 8hr State Implementation Plan modeling Scenario. To in-depth understand the complicated processes we will use a post-processing tool called pyPA (Process Analysis in Python).

This figure shows two INTERCONNECTED CYCLES, each driven by sunlight and each needing the other to function. The top boxes include mostly organic processes and the bottom boxes include mostly inorganic processes. THE INTERACTION OF THOSE TWO LEADS TO OZONE FORMATION. The consumption of VOCs and Nox occurs by a free radical in which the principal processes are RADICAL INITIATION, PROPAGATION and TERMINATION. Those radicals are initially released in the atmosphere by PHOTOLYSIS of either organic compounds (eg aldehydes) or inorganics (ozone). They then react with other VOCs, which eventually results in further formation of hydroxyl radicals. Closely coupled to this radical chain is another cycle in which new NO is first oxidized to NO2 by reacting with intermediate products that result from the OH-VOC chemistry, such as alcoperoxy radicals. The newly reacted NO2 then photolyzes to produce O3 and more NO that can once again react with the intermediate products I just mentioned and be oxidized to NO2. The O3 produced by NO2 photolysis could then lead to further formation of radicals. In this presentation, we will focus on the organic process cycle and more specifically the OH cycle and how this is driving the entire O3 chemistry. The OH cycle represents the average amount of times a hydroxyl radical propagates. The number of OH cycles multiplied by the total amount of new OH would give the amount of total OH that has reacted in the system. There are three ways to affect the OH cycles. 1. Through termination (OH reacts either with other OH to form hydrogen peroxide, H2O2, or with NO2 to form nitric acid) 2. Through increase or decrease of propagation and 3. Through increase or decrease of OH organic sources.

Sensitivity Runs CO: favors propagation of OH Xylene and Toluene: internal sources of OH, lead to formation of HCHO that when photolyzed produces OH HCHO: direct source of OH Examples of ways to affect the OH cycles are the following. CO is able to impact the propagation of radicals. Xylene and Toluene are aromatic hydrocarbons that when attacked by a radical produce molecules of formaldehyde, among others, that then is photolyzed to form more radicals. In other words, xylene and tuloene are internal sources of OH. As just mentioned, formaldehyde is a direct source of radicals. We performed sensitivity runs using these compounds.

List of Sensitivity Runs Performed Base Case (no imputation) 0.25xCO 4xCO 2xXYL_TOL FL_FORM CO_FORM This is the list of the sensitivity runs that we performed. Starting from the base case we performed the following imputations. For the 0.25xCO we reduced the area CO emissions by 3/4, for the 4xCO we increased the area CO emissions by a factor of 4, for the 2xXYL_TOL we doubled the area emissions of xylene and toluene, for the FL_FORM we imputed the emissions of 13 flares and for the CO_FORM we imputed the area formaldehyde emissions. I will now explain in more detail these imputations.

VOC Addition: CO and Xylene+Toluene Two CO sensitivity runs 0.25xCO: 1/4 times the gridded CO emissions 4xCO: 4 times the gridded CO emissions Full length run (08.22-09.06.2000) Domain wide imputation (36km, 12km, and 4km) Xylene and Toluene sensitivity run 2xXYL_TOL: 2 times the Xylene and Toluene gridded emissions Full length run Domain wide imputation As said earlier CO favors the propagation of radicals. This is why we chose to perform two different CO imputations, to observe how that impact on the OH cycle would affect the formation of O3. For the 0.25xCO we reduced the gridded CO emissions by a factor of 4. For the 4xCO case we increased the gridded CO emissions four times. For both of these imputations we performed a run for the entire 2000 episode from the 22nd of August to the SIXTH of September for the entire 36, 12 and 4km domain. To observe the impact on O3 formation due to higher concentrations of internal sources of radicals, xylene and toluene, we increased by a factor of 2 the area emissions of these two organic compounds for the entire episode and the entire domain.

VOC Addition: Formaldehyde Two Formaldehyde sensitivity runs Flares (FL_FORM) We assumed that HCHO emitted was equal to 1% of total VOC flow rate. Full length run Imputed 13 flare emissions Mobile sources (CO_FORM) Recent data (SWRI, 2005*) on Heavy Duty Diesel show that HCHO is 23% of VOC. HCHO was 5% of CO. We added HCHO at 4% of gridded CO emissions. 4km domain wide imputation We performed two formaldehyde sensitivity runs. For the first run we imputed flare emissions (that is point source emissions) on the east side of houston where the petrochemical facilities are (and I’ll show a map of this later on). After identifying the flares that emit the highest concentration of formaldehyde we imputed these emissions to equal 1% of the total VOC flow rate of these flares. *** ASK WILL *** This imputation was performed for 13 flares for the entire episode. Recent data provided by a 2005 project of the south west research institute on diesel exhaust show that 23% of the VOCs emitted by heavy duty diesel exhausts is formaldehyde. This amount of formaldehyde is equal to 5% of the CO emitted by mobile sources. This is the reason we chose to equalize the formaldehyde area emissions to 4% of the CO area emissions. This imputation was run for the entire episode for the 4km domain. *** “This is really just a sensitivity test. We are not suggesting that this is what actually is missing from the inventory. What we are doing is we are just testing the sensitivity of the model to changes in the radical system to see how that affects O3 formation.”*** *Reference: Diesel Exhaust Standard-Phase II: CRC Project No. AVFL-10b SwRI Poject No. 03.10410 Fanick, Robert. 2005

Process Analysis: Application CAMx/CMAQ quantifies the chemical and physical processes that lead to ozone formation for each grid cell Process Analysis is a post processor suite of programs that extracts, aggregates and calculates several chemical parameters Chemical parameters Total OH Balance Total VOC Reacted New NO Total NO Reacted Total NO2 Available O3 Source Balance O3 Balance New O3 per new NO and VOC reacted The tool used to quantify these processes is process analysis in python (pypa), described earlier today by Barron Henderson. PyPa is really useful to quantify chemical and physical processes that lead to O3 formation for each grid cell or for any focus area of interest. It is a post-processor suite of programs combined to extract, aggregate and calculate several chemical parameters. These parameters include the total OH balance, for which I have talked earlier and I will talk later on as well. The total VOCs reacted, that shows how much of the OH reacted with VOCs, blah… blah… blah… and the new O3 per new NO and VOC reacted. Those two ratios help us understand where O3 came from.

08/25 Base O3 12 13 14 15

HOUSTON

VOC sensitivity: C35C O3 Impact SCENARIO Base 0.25xCO 4xCO 2xXYL_TOL FL_FORM CO_FORM O3 CHEMICALLY PRODUCED 131.9 126.11 148.7 153.81 153.28 167.78

Conclusion pyPA provides quantification of important physical and chemical parameters that help us understand and explain how O3 is formed. We have shown by this quantification the impacts of VOC sensitivity analyses to physical and chemical parameters of the O3 chemistry.

Acknowledgements Funding Special thanks Eight Hour Ozone Coalition HARC H60 “Regional Transport Modeling for East Texas” Jay Olaguer, Project Officer Special thanks Jim Smith, TCEQ for supplying CAMx ready files for base1b 8-H SIP Case UNC MAQ Lab Houston advanced research center