BARRY LEFER UNIVERSITY OF HOUSTON DEPARTMENT OF EARTH AND ATMOSPHERIC SCIENCES HOUSTON, TX

Transcription

1 DALLAS MEASUREMENT OF OZONE PRODUCTION SENSOR (MOPS) AQRP PROJECT FINAL REPORT SUBMITTED TO UNIVERSITY OF TEXAS - AUSTIN AIR QUALITY RESEARCH PROGRAM (AQRP) BY: BARRY LEFER UNIVERSITY OF HOUSTON DEPARTMENT OF EARTH AND ATMOSPHERIC SCIENCES HOUSTON, TX WILLIAM BRUNE THE PENNSYLVANIA STATE UNIVERSITY DEPARTMENT OF METEOROLOGY UNIVERSITY PARK, PA DECEMBER 1, 2011

2 Executive Summary Two Measurement of Ozone Production Sensors (MOPS) were deployed in the Ft. Worth area and made ozone production rate measurements during the period from August to October This project basically concerned the development and deployment of the MOPS. The delay in receiving funding meant that the MOPS could be developed and deployed, but that there was little to no time to do proper testing and modifications of the MOPS before they were deployed. None-the-less, we expect useful results will be retrieved from both MOPS and we intend to submit a follow-on technical report once all of the laboratory tests and analyses are completed. Retrieving the ozone production rates for Eagle Mountain Lake (C75) is more difficult than for Meacham (C13) because of technical issues. However, once these technical issues are overcome and the tests and validations are complete, the results from both MOPS will be provided to TCEQ for evaluation and use. The validation and analysis of these MOPS results will provide the basis for an M.S. thesis of a student who has been involved in the development. Several results are notable, although preliminary until further tests are completed. First, at Meacham (C13), ozone production is significant in mid-morning, with ozone production rates having peak values of ppbv hr -1 on sunny days with the wind from the south. On cloudy days, the measured ozone production was less than 10 ppbv hr -1. The ozone production rates roughly correlate with the peak ozone that occurs later in the afternoon. Interestingly, the ozone production accumulated for each day generally exceeds the observed ozone by a factor of 1.5 to 6, suggesting that Meacham may be an ozone source region. The preliminary ozone production rate (P(O 3 )) at Eagle Mountain Lake appears to be less than at Meacham, peaking in the late morning and only exceeded 40 ppbv hr -1 a few times. These data, once more completely analyzed and validated, will challenge air quality models and provide information about the sources of ozone precursors at both Eagle Mountain Lake and Meacham Field. Future work includes the development of a technical report and data products. At this time we are unable to provide an estimate of when the data will be in their final form. The time periods of good data are as follows: Eagle Mountain Lake: September 12-15, 20-22, 29-30, October 1-7, 10-12, Meacham: September 15-16, 17-29, October 2-8, University of Houston - Department of Earth & Atmospheric Sciences Page 2 of 27

3 TABLE OF CONTENTS Section Page 1. INTRODUCTION Background NEW MOPS INSTRUMENT DESIGN (GENERATION 2.0) Primary Structure Chamber Design Photolysis Cell Design MOPS Chamber Cover System Integration and Automation MOPS System Input/Output System Control MOPS-II INSTRUMENT TESTING Wall loss testing of new chamber design NO 2 photolysis conversion testing MOPS-II INSTRUMENT DEPLOYMENT Results from Eagle Mountain Lake (C75) Results from Fort Worth Northwest (C13) Ozone production as function of NO levels, wind direction, etc Modeled vs. Measured P(O 3 ) QUALITY METRICS PROJECT SUMMARY...25 University of Houston - Department of Earth & Atmospheric Sciences Page 3 of 27

4 LIST OF FIGURES Figure Page Figure 1-1. Schematic of MOPS. Equal air flows pass through the two chambers exposed to sunlight. The sample chamber passes solar ultraviolet light while the reference cell has a film covering that blocks it. NO 2 converter cells enable the detection of NO 2 +O 3 by the dualchannel UV-absorption O 3 monitor....7 Figure 2-1. MOPS system flow diagram....8 Figure 2-2. MOPS system. A conceptual drawing of the system with the cover panels removed. The chamber cover is shown in the open position....9 Figure 2-3. Photolysis cell design (left) and completed unit (right). The Phoseon UV Lamp is mounted on top of the cell Figure 2-4. The new MOPS zeroing cover in the closed (left) and open positions (right). The cycling of this UV absorbing cover every 10 minutes has proven to be the best way to achieve an effective instrument zero Figure 2-5. MOPS system I/O diagram Figure 2-6. MOPS system interface...14 Figure 4-1. Installation of the MOPS at Eagle Mountain Lake was performed during the week of August 1 st, 2011 and removed October 22 nd, Figure 4-2. Results of P(O 3 ) sensitivity to NO 2 additions at EML on August 20-22, Figure 4-3. Installation of the MOPS at Ft. Worth Northwest, Meacham (C75) was performed the week of August 15 th, 2011 and was removed October 23 rd, Figure 4-4. Preliminary ozone production rate at Ft. Worth Meacham (C13). Left: ten days of ozone production (blue), ozone (red), and nitric oxide (black). Right: two days of ozone production (blue), ozone (red), nitric oxide (NO), and sunlight (green) Figure 4-5. Preliminary ozone production at Ft. Worth Meacham (C13). The ozone production rate (black), ozone (red), and cumulative ozone produced during each day (blue) indicate days with different levels of ozone production University of Houston - Department of Earth & Atmospheric Sciences Page 4 of 27

5 1. INTRODUCTION 1.1 Background The Dallas-Fort Worth-Arlington Metroplex (DFW) includes approximately 6.5 million people, making it the largest metropolitan area in Texas and the 4th largest in the United States. Given that the DFW area does not include large petrochemical facilities; the primary source of the ozone precursor NOx and VOCs emissions are the significant mobile source emissions and a number of large point sources, specifically electric power plants and cement kilns. While the ozone design value for DFW is very close to being in compliance with NAAQS 8-hr ozone standard of 84 ppbv it is interesting to note that ozone levels have not decreased significantly in recent years. Even in an environment where NOx and VOC emissions from automobiles have been gradually decreasing (Parrish, 2006). In addition, improvements in the production of natural gas from a combination of horizontal drilling and hydraulic fracturing of the Fort Worth Basin of the Barnett Shale formation have resulted in a dramatic increase in both number natural gas wells and production of natural gas in the DFW region. The network of 18 TCEQ ozone monitoring sites in the DFW area is designed to capture both upwind and downwind ozone mixing ratios, interestingly, the peak ozone values are frequently observed along the northwestern border of the network. This is also a region with a high concentration of natural gas wells. These recent developments have opened the door to speculation regarding the relationship between ozone levels and the natural gas activities in DFW area. The ozone budget, equation (1), shows that ozone production depends on the ambient air chemistry, surface deposition, and local meteorology. [ O3 ] v [ O3 ] po3 lo3 [ O3 ] ui t H xi (1) P(O 3 ) SD A P(O 3 ) is net chemical production consisting of chemical ozone production, p O3 and chemical loss, l O3, SD is surface deposition consisting of the deposition velocity, v, divided by the mixed layer height, H, times the ozone concentration, and A is advection in the horizontal and vertical. The ozone budget it typically tested by comparing the measured ozone to ozone calculated by an airshed model that contains all that is known about the meteorology, emissions, and the atmospheric chemistry. Each one of these terms has significant uncertainty, resulting in uncertainties in the model s ability to provide guidance for regulatory action and to assess the efficacy of such actions. Since regulatory actions address the emissions of gases that lead to ozone production, the most important term to test in equation 1 is the net chemical production term, P(O 3 ). All the other terms are proportional to ozone, so that ozone will decrease if the net chemical production decreases and will increase if the net chemical production increases. We have successfully deployed both the measurement of ozone production sensor (MOPS) instrument to continuously measure ozone production during the 2009 SHARP and 2010 CalNex campaigns in Houston and California, respectively. To help further investigate the behavior of ozone in Dallas, and the possibility that the new gas wells will lead to increases in ozone, we built and operated two new instruments that directly measured the ozone production rate (Cazorla and Brune, 2010) at two TCEQ University of Houston - Department of Earth & Atmospheric Sciences Page 5 of 27

6 monitoring sites during a three month period starting in August These instruments are described in more detail below. 1.2 Description of the Measurement of Ozone Production Sensor The MOPS has three components: two environmental chambers continuously exposed to solar radiation, a nitrogen dioxide-to-ozone conversion unit, and a modified ozone analyzer. A schematic of the instrument is shown in Figure 1-1. The sample chamber s Teflon walls transmit solar ultraviolet light so that the air in the sample chamber undergoes the same photochemistry that takes place in the ambient air. The reference chamber has a film that blocks radiation of wavelengths less than 400 nm. As a result, the reference chamber limits the production of hydroxyl radicals (OH) generated by the photolysis of ozone followed by the reaction with water vapor. The photolysis of nitrous acid (HONO), a source of OH radicals, is also constrained. Similarly, the film on the reference chamber restricts the production of hydroperoxy radicals (HO 2 ) produced by the photolysis of formaldehyde (HCHO). With radical chemistry eliminated, the only ozone in this chamber comes from the photostationarysteady state (PSS) of the species NO, NO 2, and O 3. Since it is not possible to eliminate radical production without affecting NO 2 photolysis near 400 nm, the PSS in the reference chamber tends to shift O 3 toward NO 2. The total amount of ozone in the reference chamber, therefore, is conserved in the form of NO 2 plus O 3. Some of the ozone produced in the sample chamber reacts with ambient NO and is partitioned into NO 2 according to the NO x PSS. At the same time, differences in the NO 2 photolysis in the two chambers could cause the partitioning of ozone and NO 2 in the two chambers to be different. The difference, nevertheless, between the total sum NO 2 +O 3 present in the sample chamber minus the sum in the reference chamber cancels out the PSS component of ozone production and is only the component associated with the production of new ozone by radical chemistry. University of Houston - Department of Earth & Atmospheric Sciences Page 6 of 27

7 Figure 1-1. Schematic of MOPS. Equal air flows pass through the two chambers exposed to sunlight. The sample chamber passes solar ultraviolet light while the reference cell has a film covering that blocks it. NO 2 converter cells enable the detection of NO 2 +O 3 by the dualchannel UV-absorption O 3 monitor. The strategy, therefore, to is to determine the differential of the sum O 3 +NO 2 between the two chambers and divide it by the exposure time, τ, of the air inside them by means of equation (2) to determine the ozone production rate. P(O 3 ) = ΔO 3 / τ (2) P(O 3 ) is the net chemical ozone production, ΔO 3 is the difference in O 3 +NO 2 between the sample and reference chamber after the NO 2 has been converted into O 3. The detection limit for the first generation MOPS the instrument was 0.67 ppb/h for the 10-minute average data. This detection limit was lowered by improving the sensitivity of the detection cells. A better sensitivity of detection enables the use of smaller sample and reference chambers and shorter exposure times. At the 95% confidence level, the absolute uncertainty of the first generation MOPS instrument was 30%. The only interference that has been identified with the MOPS technique is the uptake by NO 2, and in some cases O 3, by the walls of the sample chamber (Cazorla and Brune 2010). The problem occurred at relative humidity above 50% in the version of the MOPS chambers that were used in Houston in Since then considerable effort has gone into redesigning the MOPS chambers to suppress wall effects. Further tests are needed to confirm preliminary results that the MOPS chambers used in Ft. Worth did in fact have lower wall effects, but preliminary results from laboratory tests indicate that the new MOPS chambers do not have the 50% relative humidity restriction. Previous research on Texas air quality focused on evaluating key parts of photochemical ozone production to inform the SIP modelers on areas such as the OH radical budget (Mao et al., 2010), comparisons of different chemical mechanisms (Chen et al., 2010), or the impact of clouds and aerosols on photochemistry (Flynn et al., 2010). A goal of the MOPS project was to examine net ozone production in order to directly answer the question of whether the natural gas activities around DFW are contributing to the local ozone problem. The preliminary analysis is not yet able to answer this question although further analysis of these data may yet yield an answer. University of Houston - Department of Earth & Atmospheric Sciences Page 7 of 27

8 2. NEW MOPS INSTRUMENT DESIGN (GENERATION 2.0) For this study, the MOPS was redesigned to provide autonomous measurements that are not as sensitive to possible wall effects identified in Cazorla and Brune (2010) and are less sensitive to zero drift in the ozone sensor. The MOPS the layout of the various internal and external sensors used to monitor the MOPS performance is shown in Figure 2-1 below. Figure 2-1. MOPS system flow diagram. 2.1 Primary Structure The primary structure serves two purposes: house the system control and measurement components and support the sample and reference chambers (Figure 2-2). The lower portion of the structure houses the control and measurement components and provides a platform for the chamber structure. The chamber structure, constructed of lightweight extruded aluminum profiles with integrated t- slots., provides support for the lightweight chambers and a mount for the cover. This approach provides a rigid structure that allows for simple construction and design changes. University of Houston - Department of Earth & Atmospheric Sciences Page 8 of 27

9 Figure 2-2. MOPS system. A conceptual drawing of the system with the cover panels removed. The chamber cover is shown in the open position. 2.2 Chamber Design The chambers consist of stainless steel end pans coated with Silconert (Silcotex, Inc.) and UV-transmitting Teflon FEP film for the body of the chambers. The chambers are designed to minimize the potential for wall influence on the air sampled at the central axis of the chamber. The goal is to keep air that has been influenced by the walls near the walls, where it can be removed by a pump that is separate from the ozone sampling system. If the chamber is operating correctly, only the air that comes in the chamber along its central axis is sampled at the far end through a collector that is in the center. The dimensions of the chamber are roughly 15 cm high by 30 cm wide at the entrance end and 15 cm high by 15 cm wide at the sampling end. The total volume is about 15 liters. Air that enters through the ½ inlet is redirected through a cap that has holes drilled in the sides. It is spread rapidly behind a stainless steel screen and then enters the chamber already spread across the chamber area. All surfaces at the entrance are coated with SilcoNert 2000 to inhibit loss of HO x precursors. After the air passes through the screen, it passes through horizontal wires spaced 1 cm apart. More electrical current flows through the wires at the top than at the bottom so that the air at the top is heated ~2 o C above the air at the bottom, which is roughly at ambient temperature. This temperature gradient creates vertical stability. In the horizontal direction, the air is accelerated as the cross sectional area decreases, creating a small acceleration that inhibits meandering of the flow. As the air comes to the sampling end, the flow rate going into the ozone sampling inlet divided by the flow rate going to the external pump is proportional to the areas of the sampling cone to the outer area. As a result, there are no sideways forces and the flows then follow streamlines with the flows in the middle being sampled by the sampling cone that leads to the ozone instrument. Smoke tests and tests with a variable ozone source gave results consistent University of Houston - Department of Earth & Atmospheric Sciences Page 9 of 27

10 with wall-free sampling. Additional laboratory testing over the next months will demonstrate this improved flow characteristic quantitatively. 2.3 Photolysis Cell Design The photolysis cell is composed of two Teflon end caps that sandwich a pair of 38mm OD quartz tubes. A Phoseon Starfire UV lamp is mounted on top of the cell and the other three sides are first-surface mirrors to reflect the UV radiation. The intensity of the UV lamp is user controlled via the Labview interface. Figure 2-3. Photolysis cell design (left) and completed unit (right). The Phoseon UV Lamp is mounted on top of the cell. This arrangement is highly efficient at converting NO 2 to NO + O 3. Even when the initial NO 2 is greater than 50 ppbv, the conversion efficiency is more than 85%. Two cells one for the sample flow and one for the reference flow are both exposed to the same UV light. Although the cell with the sample is closer to the light source than the cell with the reference flow, two tests show that the conversion efficiency in the two cells is essentially identical. The use of first-surface mirrored walls causes both cells to be exposed to the same amount of UV. University of Houston - Department of Earth & Atmospheric Sciences Page 10 of 27

11 2.4 MOPS Chamber Cover During the initial instrument tests prior to deployment, it became clear that the zero/baseline tests described in the QAPP would not be sufficient and a different approach to determining the baseline for the MOPS was necessary. Tests late in the development process just prior to deployment in Ft. Worth showed that adding zero air to the chamber to get the instrument baseline as proposed in the QAPP did not give a good baseline. The addition of zero air did cause the ozone signal, which is the difference between the sample and the reference chambers, to go to zero. The ozone level in this zero air was 0 ppbv according to a measurement by a second, calibrated ozone sensor. However, because the zero air addition changed the relative temperatures of the air sampled by the ozone instrument, this zero was not the true instrument baseline. Instead, a new, more comprehensive method of determining the instrument baseline was quickly devised and deployed. It relied on covering both chambers with a Lexan cover (Figure 2-4), thus cutting off the actinic UV light to both chambers, thereby eliminating O 3 production in the sample chamber. Although this difference signal was not always zero due to a baseline drift with an unknown cause, it was the true zero because no ozone production was occurring in either chamber when they both were covered with the Lexan. Laboratory tests show that covering the chambers with Lexan completely suppresses ozone production. By periodically covering and uncovering the two chambers with the Lexan cover, the difference signal with the cover open minus the difference signal with the cover closed (i.e., the baseline) gave the ozone signal from which the ozone production rate could be determined. Since the baseline drifted by as much as 5 ppbv during each day, the baseline needed to be checked frequently. The cover s cycle had a period of 16 minutes, 8 minutes open and 8 minutes closed. This period was chosen to allow for the chamber s 4 minute residence time as determined by tests in which a mercury lamp was placed above the chambers and produced ozone by 185 nm photolysis. Opening and closing the cover gave a clear ozone signal. This method did provide an excellent baseline in the laboratory, but in Ft. Worth, small temperature shifts between the sample and reference chambers shifted the baseline of the ozone sensor by 1-3 ppbv as a function of temperature. This shift will take about a year to quantify and then refine the data. Thus, periodically covering both chambers appears to give the best baseline of the MOPS and provided a reasonable balance between the need to collect measurements and track changes in the instrument baseline as a function of changing temperatures. This procedure requires mechanical arms to periodically move a cover over both chambers. The short amount of available development time caused us to adopt a quick solution that will be improved upon for future deployments. The motor and electronics that have been used for decades on acid deposition samplers were adapted for use on the MOPS. In acid deposition sampling, this system enabled the sampling of wet deposition to be separate from the sampling of dry deposition by moving a cover from one bucket to another when it was raining. We found a way to control this movement electronically, but the cover also moves over the chambers when it is raining, thus protecting the chambers as well as providing a periodic instrument baseline check. The initial mechanism for the cover was probably the most unreliable part of MOPS because the cover sometimes got stuck in the open position and could not close. We University of Houston - Department of Earth & Atmospheric Sciences Page 11 of 27

12 designed, built, and tested a new mechanism in late August and early September and installed it on both MOPS in mid September. This mechanism worked perfectly on the MOPS at the Meacham Field site until the MOPS were retrieved in late October; the mechanism on the MOPS at Eagle Mountain Lake worked well until an arm came loose in late October. This new cover mechanism enabled both MOPS to measure the ozone production rate for several weeks unattended. It has been continually running in the laboratory since the end of the field phase without any problems. Figure 2-4. The new MOPS zeroing (baseline check) cover in the closed (left) and open positions (right). The cycling of this UV absorbing cover every 10 minutes has proven to be the best way to determine an effective instrument baseline. 2.5 System Integration and Automation MOPS System Input/Output Figure 2-5 shows the input/output diagram for the MOPS system. A small form factor PC running Microsoft Windows 7 and National Instruments (NI) Labview is used as a central control PC. It provides control for the NI multifunction data acquisition hardware (USB-6343) and the National Control Devices relay controller (NCD R45PL_USB). It also provides communication with the University of Colorado ozone instrument, which is smaller, uses less power, and incorporates many more internal sensors to monitor the performance of the instrument, unlike the available commercial instruments. The NI USB-6343 is the I/O hub for measuring the output from various sensors and producing input for several components. University of Houston - Department of Earth & Atmospheric Sciences Page 12 of 27

13 Figure 2-5. MOPS system I/O diagram. University of Houston - Department of Earth & Atmospheric Sciences Page 13 of 27

14 System Control Labview was chosen as the measurement and control software because of developing control interfaces and hardware communication is relatively easy. National Instruments offers a wide range of data acquisition hardware that integrates seamlessly with Labview. Using LogMeIn and a cellular modem allows for the system to be controlled remotely. A screenshot of the software interface is shown in Figure 2-6. Figure 2-6. MOPS system interface The user sets various parameters via the interface and can also monitor the sensor and instrument input. Overall, 60 parameters are stored in a data file every second. The extensive monitoring of temperatures and other instrument behavior allows us to diagnose problems remotely, sometimes being able to repair them remotely. It also enables us to understand subtle temperature effects on the ozone sensor measurements because we can replicate these temperature measurements in the laboratory and diagnose the impact on the ozone measurements. This next-generation MOPS is much more sophisticated and robust than the first version. Much thought has gone into automating it, remotely monitoring its performance, and enabling it to run without more than weekly maintenance. Deploying one MOPS at Eagle Mountain Lake (C75) for about nine weeks and another at Ft. Worth Meacham (C13) for about seven weeks not only generated weeks of ozone production rate measurements but also taught us how to improve the MOPS so that it can make high quality measurements while essentially being unattended. University of Houston - Department of Earth & Atmospheric Sciences Page 14 of 27

15 When the MOPS deployment ended in late October, the MOPS were brought back to Penn State for laboratory tests and calibrations. These tests will enable us to calibrate the ozone production rate measurements, understand possible temperature effects on the measurement, including the ozone sensor, and understand any possible wall effects and correct for them. It was truly unfortunate that we did not received funding until six weeks before we had to deploy and thus had time only to build and minimally test the MOPS. Both MOPS have been undergoing strenuous laboratory testing since November. Until we have the results of these tests, we will not know how reliable the results from this deployment are. University of Houston - Department of Earth & Atmospheric Sciences Page 15 of 27

16 3. MOPS-II INSTRUMENT TESTING 3.1 Wall loss testing of new chamber design Because of the tight time constraints imposed by contract and logistical issues, extensive laboratory testing is currently progress and will be completed after the submission of this Final Report. The initial pre-deployment smoke tests and tests with a variable ozone source gave results consistent with wall-free sampling. These smoke tests and ozone point source tests can give only qualitative results. Qualitatively, the smoke entered the chamber and formed small circulations with diameters of a few centimeters, thus mixing with nearby air. As the smoke accelerated due to the wedge shape of the chamber, puffs of smoke became elongated and were accelerated toward the sampling end of the chamber. These strands of smoke followed welldefined trajectories and no longer mixed. Those nearest the center were sampled by the sampling cone in the center. Those nearest the chamber walls were pulled into the bypass flow. Other more quantitative tests have been performed. First, ozone loss was found to be less than 2% in a dry system and less that 15% when the relative humidity was 60%. This increased loss with increasing relative humidity may have occurred in the sampling line where the ozone and humidified air were mixed and not in the chamber. Second, Teflon walls tend to get saturated with nitrogen oxide compounds, so a good test of wall effects is to add CO or a VOC in the presence of real or simulated sunlight and see if ozone is produced, since the walls could provide the needed NO. No ozone production was observed. These two tests indicate that the wall effects in the MOPS chambers had an effect less than 15% on the ozone signal. More tests are underway to further quantify wall effects. 3.2 NO 2 photolysis conversion testing The NO 2 photolysis conversion testing, which was completed before the MOPS were deployed, showed that NO 2 was close to the theoretical maximum amount. This conversion is 90±3% for NO 2 less than 10 ppbv and is 80±5% for NO 2 greater than 10 ppbv. This conversion efficiency is identical for both the sample and the reference lines. University of Houston - Department of Earth & Atmospheric Sciences Page 16 of 27

17 4. MOPS-II INSTRUMENT DEPLOYMENT One of the challenges of this project was to design the new MOPS in such a way that it could run unattended for extended periods of time, as opposed to the continual maintenance and monitoring that earlier versions required. Unattended operation and low maintenance needs means that the new generation is better suited for long-term monitoring instead of short intensive measurement campaigns. As with all new instrument development and initial deployments, some difficulties were encountered. Prior to deployment the University of Colorado instruments were damaged in the lab and there was not time to repair them and still meet the deployment schedule. Instead, commercial Thermo 49c and 49i instruments were modified in order for them to be used as replacements. These Thermo instruments were modified by removing the internal ozone scrubber and having Cell A draw air from the sample chamber and Cell B draw air from the reference chamber. The instrument solenoids remained in use and switched the flow between the two cells, negating any differences that may exist between the two cells as it would be in normal operation. In this configuration the O 3 measurement reported by the instrument was the difference in O 3 between the sample and reference chambers. Additional insulation was added to the internal tubing and measurement bench to reduce changes in temperature. In general, these instruments worked well, however their use required additional cooling and provided fewer operational parameters than what the University of Colorado instruments were designed to provide. The primary limitation with the Thermo instruments was that they only reported one temperature and pressure for both cells, while the University of Colorado instruments report the pressure and temperature of each cell independently. The reduced information from the commercial instruments has complicated the data analysis somewhat since there were some temperature differences between the sample and reference chambers, and a slight temperature shift when the cover was closed, leading to the need for more extensive post-deployment testing, however the data is still sufficient to address the objectives and will be further refined after the additional testing is completed. The initial MOPS deployment to Eagle Mountain Lake (C75) (EML) went smoothly and the instrument operated well for several days while testing and verification checks were being performed. Remote communication was possible by the use of a cellular modem installed on the internal MOPS PC and services through LogMeIn.com. Although the communication software for the modem was set to always maintain an active connection to the internet, connectivity problems arose after the Penn State and UH personnel left the site. To help maintain connectivity, a local college student was hired to perform routine checks at the site and to be on-call as the teams at Penn State and UH discovered issues that couldn t be resolved remotely. On the next trip to Ft. Worth to deploy the MOPS at Ft. Worth Northwest, Meacham (C13) a new router and network power switch were installed in both MOPS instruments. The router s software was configured to reboot the cellular modem if signal was lost. On the occasions that rebooting the modem did not correct the connectivity issues, the network power switch, which was actively pinging an off-site server, would cycle the power to the router and modem after a preset period. This dramatically increased the availability of the MOPS instrument for remote operation. Occasionally, the MOPS PC would lose its connection to the router, also rendering it unavailable for remote access. The solution to this problem was to have the computer send a heartbeat signal to the network power switch. When the PC network University of Houston - Department of Earth & Atmospheric Sciences Page 17 of 27

18 connection was lost, the heartbeat would no longer be received by the power switch, and after a set period it would cycle the power on the PC. Since the PC was continuously writing data and set to automatically restart and launch all programs when power was restored, instrument downtime was kept to a minimum. Aside from the problems mentioned previously with the chamber cover, the only other significant recurring problem was from occasional power loss to the MOPS at EML. This occurred a few times during the deployment, but only happened when heavy rain caused the GFCI outlet the instrument was plugged into tripped. It is still unclear as to why the outlet tripped since the in-use weather cover was securely closed, but power was typically restored within 24 hours by the local student. 4.1 Results from Eagle Mountain Lake (C75) Figure 4-1. Installation of the MOPS at Eagle Mountain Lake was performed during the week of August 1 st, 2011 and removed October 22 nd, Because of the tight timelines, compounded by delays in contracting and funding, and extended deployment into October, the analysis of the EML data is in progress and will take several more months to complete. This is primarily due to the ongoing testing at Penn State so that the data will be further refined. Preliminary P(O 3 ) at Eagle Mountain Lake appears to be less than at Meacham, peaking in the late morning and only exceeded 40 ppbv hr -1 a few times. Technical issues need to be resolved in laboratory studies before more definitive values can be determined for this location. However, P(O 3 ) sensitivity to added NO 2 (Figure 4-2) showed a consistent pattern: P(O 3 ) increased as the amount of added NO 2 increased. A small NO 2 flow was added to the ambient flow just upstream of the chambers (20-22 August). There are two pieces of evidence that this P(O 3 ) is real. First, in a nighttime NO 2 addition test, P(O 3 ) stayed close to zero, and second, enhanced P(O 3 ) was observed whether the MOPS was measuring O 3 only or O 3 + NO 2 (NO 2 converted to O 3 by the photolysis cell between the chambers and the ozone sensor). University of Houston - Department of Earth & Atmospheric Sciences Page 18 of 27

19 120 MOPS1 - Eagle Mountain Lake - doy: PO 3 ppbv/hr added NO x ppbv Figure 4-2. Results of P(O 3 ) sensitivity to NO 2 additions at EML on August 20-22, University of Houston - Department of Earth & Atmospheric Sciences Page 19 of 27

20 4.2 Results from Fort Worth Northwest (C13) Figure 4-3. Installation of the MOPS at Ft. Worth Northwest, Meacham (C75) was performed the week of August 15 th, 2011 and was removed October 23 rd, Ten days of preliminary ozone production rate measurements show the promise of the MOPS (Figure 4-4). The first finding is that the days of highest ozone production are the same as the days of highest ozone and highest morning NO (Figure 4-4 (a)). On two of those days, the sequence of events is increasing NO at morning rush hour, the simultaneous increase in ozone production and incoming solar energy, followed by the rise of ozone. University of Houston - Department of Earth & Atmospheric Sciences Page 20 of 27

21 Figure 4-4. Preliminary ozone production rate at Ft. Worth Meacham (C13). Left: ten days of ozone production (blue), ozone (red), and nitric oxide (black). Right: two days of ozone production (blue), ozone (red), nitric oxide (NO), and sunlight (green). The ozone production rate can be accumulated during each day to give the cumulative ozone production for each day ( Figure 4-5). The preliminary conclusion is that much more ozone is produced at this site than is observed, sometimes by as much as a factor of six. If the preliminary P(O 3 ) measurements are proven to be accurate by the laboratory tests, this observation suggests that the Ft. Worth Meacham site is in the midst of an ozone production region and that much of the ozone produced there is exported elsewhere. This analysis will be extended to all days at both sites for which the ozone production measurements are found to be of sufficient quality. University of Houston - Department of Earth & Atmospheric Sciences Page 21 of 27

22 Figure 4-5. Preliminary ozone production at Ft. Worth Meacham (C13). The ozone production rate (black), ozone (red), and cumulative ozone produced during each day (blue) indicate days with different levels of ozone production. 4.3 Ozone production as function of NO levels, wind direction, etc. Analysis is in progress and will be determined from deployment results. However, an indication of the ozone production as a function of added NO 2 is given in Figure 4-2. The data are from the MOPS at Eagle Mountain Lake (C75). 4.4 Modeled vs. Measured P(O 3 ) After all the laboratory instrument characterizations have been completed and revised P(O 3 ) values have been calculated, that can be compared to calculated P(O 3 ) determined by a 0-D photochemical box modeling using measurements of NO, NO 2, solar radiation (scaled to TUV radiative transfer model output) and VOCs from the co-located auto-gc instruments as inputs. This P(O 3 ) measured vs. photochemical modeling analysis task is part of a M.S. thesis project that is currently in progress and will be provided to the AQRP as a technical report once the analyses are complete (approximately months). University of Houston - Department of Earth & Atmospheric Sciences Page 22 of 27

23 5. QUALITY METRICS Since the MOPS is a new instrument, there were no existing standards for its operation. The results from these field experiments will help establish these standards for future deployments. Preliminary quantitative acceptance criteria include the following: ozone production rate numbers are zero or positive for one-hour averages; limit-of-detection of 5 ppbv hr -1 or less for one-hour averages, as determined by standard deviation of the mean of the individual few-minute measurements. Experience on three previous field campaigns suggests that the following criteria currently apply. QA/QC Check Frequency Acceptance Criteria Corrective Action Chamber Temp & RH Hourly Ambient ±20 C Check for inlet for proper flow Check chamber position for other site factors Check for inlet for proper flow Chamber Temp & RH Measurement system contribution zero check Measurement system contribution O 3 linearity/precision Measurement system contribution chamber O 3 transmission Hourly Several times per hour Multipoint O 3 challenges prior to beginning and after end of sampling Weekly Ambient absolute humidity ±5% O 3 channels read below 5 ppbv when the cover is closed Linear fit of or better Both chambers within 15% of each other Within 20% of nearby O 3 monitors under conditions when concentrations should be expected to be relatively uniform Check chamber position for other site factors Check all lines for leaks Check ozone instrument Check ozone instrument with NO 2 converters Check remaining MOPS components Clean components when source of nonzero is found. Check O 3 generator settings Recalibrate or repair Purge chambers with zero air and ozone Replace Teflon chambers University of Houston - Department of Earth & Atmospheric Sciences Page 23 of 27

24 QA/QC Check Frequency Acceptance Criteria Corrective Action Tighten fittings Chamber flow rate Weekly Nominally 1-5 LPM NO 2 conversion efficiency Chamber UV transmission Chamber HO 2 Prior to beginning and after end of sampling Prior to beginning and after end of sampling, then weekly during operation Prior to beginning and after end of sampling Yet to be determined until NO 2 conversion system is built; Previous versions were >75% conversion efficiency for NO 2 < 35 ppbv >90% Ambient ±10% Replace Teflon chamber if hole or tear is found Inspect NO 2 conversion system for leaks Clean as needed Clean chambers Replace Teflon chamber Check for inlet for proper flow Inspect instrument for leaks Clean chambers During the deployment it was necessary to adjust the QA/QC checks that were performed. The operational schedule allowed some checks to be performed frequently while others have been performed in post-deployment laboratory tests. Temperature and humidity were monitored continuously, and zero checks were made approximately every 10 minutes, much more frequently than proposed. Chamber flow rates were checked whenever someone was at the sites, which was less than once a week. Chamber ozone transmission was checked before the deployments and is being checked again in the post-deployment laboratory testing. The UV transmission, ozone linearity, and the HO 2 production in the sample and reference chambers are also being checked in the laboratory after the deployment. University of Houston - Department of Earth & Atmospheric Sciences Page 24 of 27

25 6. PROJECT SUMMARY Both Measurement of Ozone Production Sensors were deployed in the Ft. Worth area and made ozone production rate measurements during the period from August to October This project basically concerned the development and deployment of the MOPS. The delay in receiving funding meant that the MOPS could be developed and deployed, but that there was little to no time to do proper testing and modifications of the MOPS before they were deployed. None-the-less, we expect useful results will be retrieved from both MOPS and we intend to submit a technical report once all of the laboratory tests and analyses are completed. Retrieving the ozone production rates for Eagle Mountain Lake (C75) is more difficult than for Meacham (C13) because of technical issues. However, once these technical issues are overcome and the tests and validations are complete, the results from both MOPS will be provided to TCEQ for evaluation and use. The validation and analysis of these MOPS results will be provide the basis for an M.S. thesis of a student who has been involved in the development. Several results are notable, although preliminary until further tests are completed. First, at Meacham (C13), ozone production is significant in mid-morning, with ozone production rates having peak values of ppbv hr -1 on sunny days with the wind from the south. On cloudy days, the measured ozone production was less than 10 ppbv hr -1. The ozone production rates roughly correlate with the peak ozone that occurs later in the afternoon. Interestingly, the ozone production accumulated for each day generally exceeds the observed ozone, suggesting that Meacham may be an ozone source region. Preliminary P(O 3 ) at Eagle Mountain Lake appears to be less than at Meacham, peaking in the late morning and only exceeded 40 ppbv hr -1 a few times. These data, once more completely analyzed and validated, will challenge air quality models and provide information about the ozone sources at both Eagle Mountain Lake and Meacham Field. Future work includes the following report and data products: 1) Data & Technical Report - Additional testing is being conducted at Penn State, however these tests are taking longer than expected to provide the necessary results to interpret the data and generate the most useful product. At this time they are unable to provide an estimate of when the data will be in its final form. The Technical report depends on this data, so the delivery date on that is also uncertain. We will continue to work with Penn State and will provide updates to you as we receive them, however please feel free to check back with us periodically and we will ask for additional updates. 2) Time periods of good data: Eagle Mountain Lake: September 12-15, 20-22, 29-30, October 1-7, 10-12, Meacham: September 15-16, 17-29, October 2-8, ) General description of final data: Time stamp: will be in Day-of-year format, and will likely be on the order of 5-10 min intervals, for each data record. This may vary however and will be the fastest reasonable speed to generate the best data product from the raw data. P(O 3 ): in-situ ozone production rate in ppbv/hour University of Houston - Department of Earth & Atmospheric Sciences Page 25 of 27

26 NO 2 Addition: This column will indicate the amount of NO 2 that was added to the chamber inlet to conduct P(O 3 ) sensitivity tests University of Houston - Department of Earth & Atmospheric Sciences Page 26 of 27

27 7. REFERENCES Parrish, D. D. (2006), Critical evaluation of US on-road vehicle emission inventories, Atmospheric Environment, 40(13), Cazorla, M., and W. H. Brune (2010), Measurement of Ozone Production Sensor, Atmospheric Measurement Techniques, 3(3), Chen, S. A., X. R. Ren, J. Q. Mao, Z. Chen, W. H. Brune, B. Lefer, B. Rappengluck, J. Flynn, J. Olson, and J. H. Crawford (2010), A comparison of chemical mechanisms based on TRAMP-2006 field data, Atmospheric Environment, 44(33), Flynn, J., et al. (2010), Impact of clouds and aerosols on ozone production in Southeast Texas, Atmospheric Environment, 44(33), Mao, J. Q., et al. (2010), Atmospheric oxidation capacity in the summer of Houston 2006: Comparison with summer measurements in other metropolitan studies, Atmospheric Environment, 44(33), University of Houston - Department of Earth & Atmospheric Sciences Page 27 of 27