Base-case emissions for combustion stage at the WAP plant, were obtained from TCEQ’s Airs Facility Subsystem (AFS) files for modeling 2012 ozone episode. The average emissions for the four coal units of WAP plant, during June 2012 are described in Table 1, as emission rate in tonnes/hr (t/hr). Estimates for emissions reduction at the combustion stage were obtained from Kommalapati et al., (2018), and are summarized in Table 2 [19] [27] . Biomass co-firing is considered only in the coal units of WAP, which were Units 5 - 8, and were controlled by Selective Catalytic Reduction for NO_X.

Emission Processing System 3 (EPS3) consists of a series of FORTRAN modules that perform intensive data manipulations, producing an emissions inventory for photochemical modeling [26] . EPS3 was used to process the Airs Facility Subsystem (AFS) files publically available at the Texas Commission on Environmental Quality website [28] . This study used PREPNT, SPCEMS, TMPRL, PSTPNT, PIGEMS, GRDEM, MRGUAM modules to process data. The output from LCA study for combustion stage was used to modify only point sources file related to EGUs in the HGB area. All other emission files (Biogenic, area, nonroad, offroad, oil and gas, onroad, and emissions from Mexico and Canada) are retained intact and used in final module (PIGEMS and MRGUAM) of EPS3 for merging.

CAMx version 6.31 with improved halogen chemistry was used to predict air pollutant concentration in the HGB area. CAMx uses the 36 km master grid as coarsest grid and then moves to the finer grids of 12 km and 4 km. Lambert Conformal Conic map projection is used in Texas ozone modeling domains, where first true latitude, second true latitude, and central longitude are 33˚N, 45˚N and 97˚W respectively. The modeling domain for this study includes emission sources in Canada, the U.S., and Mexico. The coarse domain covers the area under Easting (−2735, 2592) and Northing (−2088, 1944) coordinate, also the grid areas are divided into 148 × 112 cells. The tx_12 km domain has 149 × 110 numbers of cells and the tx_4 km domain is divided into 191 × 218 cells. The five major data classes needed as input for CAMx modeling are emissions, meteorology, photolysis, geographic, air quality. In this study, four simulations were completed for each day, including a base case (no co-firing), a 5% co-firing case, a 10% co-firing case, and a 15% co-firing case. Each case has two scenarios: all-sources (biogenic, point, area, mobile), and point + biogenic sources. In all the cases, boundary condition, top concentration, meteorology, land use, photolysis are obtained from TCEQs Texas Air Quality Modeling Files and Information (2012 Episodes), except low-level emission and elevated-level emission files [29] . The CAMx output is in binary format, and VERDI software is used to visualize these results. Air quality impact analysis is considered at three air monitoring stations to compare the effect of co-firing. The details of the three selected sites are provided in Table 3 and Figure 1. University of Houston, Sugarland (C696) monitoring location, is selected as it is the closest monitoring station (11.02 km) to the WAP plant. Houston Bayland Park (C53) located at Harris County, and La Porte Sylvan Beach (C556); were selected due to high population density at Bayland Park and proximity to Houston Ship Channel for C556 respectively.

The maximum concentration of NO_x in June 2012 was observed at 7:00 hr on 21^st June, and 11^th June for both base and point + biogenic cases respectively. Co-firing of biomass does not affect NO_X concentration during the highest concentration hours. There are some reductions observed due to co-firing, which occur when the concentration is less than 7 ppb for the base case, as presented in Figure 2. There is no change of NO_x level for the 5% co-firing case, except at 14:00 hr on June 3^rd, where 1 ppb reduction was observed from a 3 ppb base-case concentration. On June 18^th (8:00 hr), the concentration reduced to 12 ppb from 13 ppb for the 15% co-firing case. The possible reasons for the insignificant effect due to biomass co-firing could be that WAP power plant is controlled for NO_X emissions by using selective catalytic reduction (SCR), and 90% emissions reduction was already observed from 1997 to 2010 [9] . In addition, 67% of NO_X in the HGB area came from mobile sources in 2014, and even in point source emissions, non-EGUs emitted an almost double amount of NO_x than EGUs in 2012 [30] . So, considering all-sources for CAMx modeling may have a lower effect on ambient concentrations at the monitoring stations (C696). In addition, meteorological conditions like wind speed and wind direction may dampen the effect of reduced stack emissions on changes in ambient concentrations [31] .

There is no significant decrease in NO_x concentration due to co-firing, at the Houston Bayland Park monitoring station (C53). Only two and four hours out of the 720 hours for June 2012 showed a 1 ppb reduction for the all-sources and point + biogenic case scenarios respectively. Both of these reductions were during low-concentration hours, reduction only observed for 10%, and 15% co-firing case. Onroad emissions of NO_x were 107.12 tons per day (tpd) in Harris County in 2012 which is 3 times more than EGUs emission of NO_x in the HGB area [3] . That means that C53 is dominated by mobile source emissions. Due to downwind wind pattern from the Gulf of Mexico, the NO_x emissions from a high stack of WAP can be transported to C53 location and influence the air quality depending on meteorological conditions of the day. Co-firing of biomass in WAP power plant does not have an effect on C556 station when all source considered. For point + biogenic sources, only three different hours shows reduction: The maximum reduction observed at 10:00 hr (6^th June) which was 1 ppb reduction from 4 ppb (25%); 1 ppb reduction observed from base case 25 ppb for both 10% and 15% ratios at 8:00 on the same day; and 5^th June at 9:00 hr showed 1 ppb reduction from 7 ppb in the 10% and 15% ratio cases. The dominant sources in 2012 for NO_x were On-road emissions in Harris County. Also, C556 located in the Houston Ship Channel (HSC) region where marine emissions have a higher impact than point sources from Fort Bend County. HSC is the largest complex of industrial sources of NO_x and VOC in HGB area [9] . The wind flow passes WAP power plant goes to Houston downtown mostly, so it is unlikely that wind carries NO_x to C556. But wind flow observed by Darby (2005) in higher ozone days in Houston often had offshore flow in the morning that switched to onshore flow in afternoon [32] .

This study added 14 different species (ALD₂, ALD_X, ETH, ETHA, ETOH, FORM, IOLE, ISOP, MEOH, OLE, PAR, TERP, TOL, and XYL) of VOCs in VERDI to get VOC emissions. There is no significant variation of VOCs due to co-firing in ambient concentrations at C696 station, for the all-sources scenario, as presented in Figure 3. The variation was only observed during lower concentration days, only for four hours in the whole month. The maximum reduction was observed on 14^th June (6%), which is only 1 ppb reduction from a 16 ppb base-case concentration. Considering only point + biogenic sources at C696, only six hours out of 720 hours, showed a reduction. The maximum drop was at 11:00 hr on 12^th June of 1 ppb reduction from a 9 ppb base-case, for both 10% and 15% co-firing. There is no reduction observed due to 5% co-firing, except at 13:00 hr on 2^nd June. At station C53, only one hour in the whole month showed a reduction (1ppb) in both cases for 15% co-firing. Co-firing of biomass in WAP has no effect on C556 station through the month, for both source scenario. Major contributors of VOC emission in HGB are area sources. In 2014, VOC emissions contribution from the area, mobile, and point sources are 61%, 23%, and 16% respectively [33] . C53 located far from WAP in Harris County where mobile emissions much higher than Fort Bend County. The probable reason for negligible reduction may be a wind effect which carries the plume from WAP to C53 on this particular hour. C556 is dominated by VOC emission from Houston

Ship Channel. In 2006, the VOC emission from WA Parish power plant and HSC was 0.5 tpd and 58.7 tpd respectively [9] . Temperature and wind account for 66% of the variation of VOCs from point sources [34] .

The maximum concentration during the June-2012 episode (105 ppb) was observed at 13:00 hr on 26^th June at C696 station for all-sources. The main reason behind maximum ozone on 26^th June is the high temperature and low cloud cover [35] . Biomass co-firing at WAP power plant has marginal reductions in ozone concentrations at the C696 location, and co-firing has effects on daily peak ozone concentrations as described in Figure 4. There are 23 different hours out of the 720 hours that show a reduction of ozone due to co-firing. The maximum percentage reduction (4.76%) was observed on 3^rd June, but it is during the low ozone level. There are 7 hourly periods (1^st June, 16:00 hr; 6^th June 11:00 hr; 14^th June, 11:00 hr; 15^th June, 11:00 and 13:00 hr; 18^th June, 12:00 hr; and 27^th June, 13:00 hr) that co-firing resulted in reduction of daily maximum 1-hr ozone concentrations. 1 ppb ozone reduction from peak can prevent 2 - 3 deaths per year, indicating the significance of achieving marginal reduction [36] . Time series graph of concentrations for the only point and biogenic sources are presented in Figure 5. The maximum concentration for this sub-case (68 ppb) was observed at 13:00 hr on June 27^th. Co-firing has a significant effect on air quality at C696, the maximum percentage (6.25%) reduction was observed at 11^th June (20:00 hr) for 15% co-firing, ozone reduced to 15 ppb from 16 ppb. Reduction of ozone was also detected on higher concentration days. On June 1^st (17:00 hr), 1 ppb ozone reduction for 15% co-firing (base case ozone 62 ppb) was observed, and 1 ppb decreases for both 10% and 15% co-firing cases was observed on 27^th June (12:00 hr). The ozone formation of C696 station is both NO_X and VOC sensitive and is in a transitional state. So, the reduction in O₃ at C696 may occur

due to the reduction of both NO_x and VOC from the power plant. A study reported that 50% reduction of NO_X from the HGB area has a huge impact on ozone in Houston and showed peak ozone reduced by 15% at air monitoring station located at the east of the HSC [37] . Also, previous studies found that ozone increased in areas where mobile and point sources dominate NO_x emissions and regions with lower NO titration. Lin et al., (2005) observed that the peak hourly ozone decreases by 70 ppb and 128 ppv in Houston when NO_x and VOC emissions from Texas point sources were ignored [38] . Also, 3% ozone concentration difference occurred for 0.16% point source emission difference in HGB. Our study shows that a small but significant change of ozone at C696 station can be observed for very low reduction of VOC and NO_X from point sources [39] . A report by TCEQ, noted that the peak ozone on September 8, 1993 reduced 1ppb from 187 ppb in Houston due to complete removal of NO_x emission from WAP [40] .

From the CAMx results of O₃ concentrations at C53 monitoring station for the base case and co-firing cases, percentage change in hourly ozone was analyzed and presented in Figure 6. The daily maximum occurred mostly during 11:00 to 14:00 hr, and the maximum during the month of June was observed on 26^th June (11:00 hr and 13:00 hr). Though there were some hours where ozone reduction was observed due to co-firing at WAP, overall reduction was insignificant at C53, for all-sources. The maximum concentration for the month of June (64 ppb) was observed at 15:00 hr of 9^th June by considering the point and biogenic sources. The maximum percentage reduction (3.33%) occurred at 13^th June (10:00 hr), the reduction is 1 ppb for 5%, 10%, and 15% co-firing ratios (base case concentration, 29 ppb). The changes due to 5% co-firing occurred

several times in this case (2^nd June, 10^th June, and 13^th June). Biomass co-firing has not affected ozone levels during the peak concentration hours. Houston Bayland Park is mostly sensitive to VOC. So, the reduction of ozone might occur due to the reduction of VOC. The ozone concentrations did not show any significant changes due to co-firing at C556 for all-sources case. Only one incident of reduction was observed at 13:00 hr on 2^nd June where, 1 ppb ozone reduced due to 15% co-firing, from a base case of 66 ppb. This reduction occurred on a relatively high ozone day. Only three hours out of 720 hours for the month of June showed a reduction for the point + biogenic case. The maximum percentage reduction (14%) occurred on June 1^st (0:00 hr), where actually 1ppb reduction noted from 7 ppb (base case) for all three co-firing cases. The monitoring station that is located in the HSC area is mostly affected by VOCs. C556 has VOC sensitive ozone in most of the hours in June’12. High ozone concentrations in the HSC area are attributed to high concentrations of hydrocarbons [41] . So, the reduction of ozone precursors from WAP has a negligible effect at C556. One of the reasons for observing negligible effect at C556 is due to the westerly wind flow that carries plume from WAP at night and influences ozone only during subsequent days. But if the winds are very light, it creates stagnant conditions in the local area where the plume is generated. Also, nighttime transport of NO_x reduces the concentration of NO_x. WA Parish plume of NO_x losses 36% to 44% on average in overnight transport and the maximum reduction for 12h of transport in darkness was 73%, which lead to lower ozone formation due to Parish plume in downwind [42] . The reason for this was lack of enough NO_X at plume to titrate background ozone which led to rapid nighttime oxidation of NO₂ during downwind transport during that particular simulation period.

C556 air monitoring station is located in Harris County and nearby the Houston Ship Channel. Time series graph for ozone concentration at C556 is presented in Figure 7. When all-sources were considered, the maximum concentration for June month was observed at 15:00 hr on 26^th June (99 ppb). The daily peak occurred during 12:00 to 14:00 hr on most of the days of the month, and June 20^th the peak occurred at 23:00 hr. The diurnal variations followed the same trend as other locations, lower concentrations observed in the early morning. The ozone concentrations did not show any significant changes due to co-firing. Only one incident of reduction was observed at 13:00 hr on 2^nd June where, 1ppb ozone reduced due to 15% co-firing, from a base case of 66ppb. For the point + biogenic sources scenario, the maximum concentration was observed on 26^th June (9:00 hr) (72 ppb). The peak ozone in the first 10 days of the month was observed on 5^th June and for the mid-10 days on 17^th June, the peaks are 55 ppb and 47 ppb respectively. In mid-10 days the daily peaks of ozone concentrations are relatively lower than other parts of the month. The lower concentration occurred during midnight to morning in most of the days. There is no significant reduction of concentrations at this location due to co-firing. Only three hours out of 720 hours for the month of June show reduction. The maximum percentage reduction (14%) occurred on June 1^st (0:00 hr), where actually 1ppb reduction noted from 7 ppb (base case) for all three co-firing cases. The other two hours where the reduction was observed were at 5^th June 12:00 and 13:00 hr, where 1 ppb reduction observed from 55 ppb and 50 ppb respectively.

Lower concentrations of SO₂ were observed during midnight to early morning throughout the month, except June 7. The peak SO₂ concentration (34 ppb) of June 2012 observed on 13^th June, 11:00 hr, as shown in Figure 8. The daily

maximum was observed in the mornings between 7:00 to 11:00 hr for most of the days. Ambient SO₂ concentrations at C696 air monitoring station were significantly affected due to co-firing of biomass with coal at the WAP power plant. The maximum reduction (25%) of SO₂ observed on 5^th June in the first 10 days of month, but it is in lower concentration day. Sulfur dioxide concentrations also show significant drop during higher concentration days also. On June 4, 8:00 hr the peak SO₂ in first 10 days, due to co-firing of 5%, 10%, and 15% ratio the concentration reduces by 5%, 5%, and 10% respectively (Figure 8(b)). The peak for the whole month was observed at June 13^th; concentrations show a gentle decrease in this day, which is 9% and 12% for 10% and 15% co-firing respectively. In higher concentration days, 5% co-firing does not have an impact mostly. Almost all of the higher concentration days observed a drop of concentration due

to co-firing while lowest SO₂ concentration didn’t show a frequent drop. For the point + biogenic sources, the maximum concentration (30 ppb) was observed at 11:00 hr on June 13^th. It is the same time where peak occurred due to all source case also. The daily maximum observed for all days during daytime except June 7^th where daily peak shows at 2:00 hr. The reduction of SO₂ concentration observed for almost all the high concentration hours with less number of hours counted that have a lower concentration. The reduction in peak concentration day (13^th June) is 3 ppb (32 ppb to 29 ppb) for 15% co-firing, 2 ppb for 10%, and 1 ppb for 5% co-firing.

Fossil fuel combustion is the largest source of SO₂ emissions. Power plants emit more than 70% of SO₂ in Houston, other industrial facilities contribute around 20%. TCEQ studied 7 major point source emissions sources in HGB area where WAP parish plant had highest (167.1 tpd) SO₂ emission in 2006 that was around two times greater than all other six sources [9] . Emissions from non-road sources in Fort Bend County were relatively lower (0.01 tpd) in 2012, and area sources emited 0.28 tpd, which was not significant [43] . There are no significant sources of SO₂ in nearby WAP power plant. Since C696 is nearby WAP power plant, Power plant emission has a great impact on air monitoring station in both cases. The highest SO₂ days affected most because of WAP is probably responsible for high SO₂ in this station. Also, wind direction and speed plays a vital role for highest concentration days. A study by TCEQ shows that Croquet monitoring station located at 22.4 km north-northeast of Parish is heavily affected when the wind blows towards station from WAP [27] . There are a few reduction hours observed in C556 for all-sources case. The maximum reduction (13%) observed at 12:00 hr and 11:00 hr on 5^th June and 6^th June for 15% co-firing. The other time periods where the reduction was observed are at 7:00, 8:00, 9:00 hr on June 5^th and 9:00 hr on June 6^th, all these hours indicating a 1 ppb reduction for 10% and 15% co-firing. No reduction observed for 5% co-firing. The reasons behind insignificant reduction that C556 located at HSC where ship emissions have a dominant impact. For point + biogenic case, the peak SO₂ concentration observed at 0:00 hr on June 1^st and the value is 23 ppb and 1 ppb reduction observed in that hour for 10% and 15% co-firing cases. There is no big drop in concentration overall except at 10:00 hr on 5^th June where 1 ppb dropped from 4 ppb for both 10% and 15% co-firing ratio which is a 25% reduction.; at 7:00 hr, 8:00 hr in same day 1 ppb drop observed for same co-firing cases. Also, 1 ppb reduction observed at 9:00 and 10:00 hr on 6^th June.

Figure 9 describes the percentage reductions observed in CO concentrations, due to co-firing considering all-sources. Co-firing has a negligible effect on CO concentrations at the C696 station. There is no change in emissions for the whole month, due to 5% co-firing, except on June 2 and 12 at 9:00 and 10:00 hr of 0.78% and 0.82% respectively. For 10% co-firing, there are changes in 16 hourly data points out of 720 hours, at a maximum reduction of 1.04% on 4^th June. For the 15% co-firing case, a total of 27 hourly data points show a reduction, with a maximum reduction similar to 10% case. There is no reduction observed during the peak CO concentration hours. Any reductions which were only observed during lower concentration hours would not affect the CO levels in view of public health or welfare. A comparison of the base case and co-firing cases doesn’t indicate a significant difference due to limiting the source categories to point + biogenic sources. The maximum reduction of 10.23%, on June 5^th at 15:00 hr for 15% co-firing ratio, CO reduces from 88 ppb to 79 ppb at this particular hour. The maximum reduction of 2% CO, for 5% co-firing was observed on June 1, 21:00 hr. All of the reductions occurred when CO was less than 150 ppb. C696 is the closest air monitoring station to WAP, and Fort Bend

County has lower point source emissions (WAP excluded), than Harris County. When all-sources were considered, modeling results suggest that reduction in CO levels would be significantly lower compared to the case where only point and biogenic sources are considered. This can be explained by the fact that mobile sources emit huge quantities of CO in composition to point sources, thereby magnifying any effect of co-firing at WAP on the CO levels observed at C696 station. Also, meteorological factors greatly influence pollutant concentrations, minimizing the effects of co-firing. The co-firing of biomass with coal in WAP does not have any notable effect on C53 and C556 air monitoring station in both cases. Only on 10^th June 7:00 hr, a very small reduction of 1ppb from 503 ppb, was observed for 10% and 15% co-firing. C53 is located in highly populated area which is greatly affected by mobile sources rather than electricity generation facilities and is located 27.64 km far from WAP. Friedfeld et al., (2002) reported that 90% of total CO came from mobile sources whereas only 8% is due to all stationary sources in HGB area for 1999 [44] . So, a small percentage of CO changes in WAP do not have a significant effect on C53 location. Area sources emit almost 10 times CO in Harris County than Fort Bend County [27] . C556, located in Harris County and non-road emissions of CO are around 13 times higher than Fort Bend County, where WAP is located. So, this small reduction of CO in WAP due to co-firing does not necessarily affect C556 air quality. Table 4 summarizes the number of reductions in daily maximum 1 hr concentrations of all pollutants that were observed at the C696 monitoring station, during the month of June 2012.