Technical Papers

Critical review of the building downwash algorithms in AERMOD

ABSTRACT

ABSTRACT

The only documentation on the building downwash algorithm in AERMOD (American Meteorological Society/U.S. Environmental Protection Agency Regulatory Model), referred to as PRIME (Plume Rise Model Enhancements), is found in the 2000 A&WMA journal article by Schulman, Strimaitis and Scire. Recent field and wind tunnel studies have shown that AERMOD can overpredict concentrations by factors of 2 to 8 for certain building configurations. While a wind tunnel equivalent building dimension study (EBD) can be conducted to approximately correct the overprediction bias, past field and wind tunnel studies indicate that there are notable flaws in the PRIME building downwash theory. A detailed review of the theory supported by CFD (Computational Fluid Dynamics) and wind tunnel simulations of flow over simple rectangular buildings revealed the following serious theoretical flaws: enhanced turbulence in the building wake starting at the wrong longitudinal location; constant enhanced turbulence extending up to the wake height; constant initial enhanced turbulence in the building wake (does not vary with roughness or stability); discontinuities in the streamline calculations; and no method to account for streamlined or porous structures.

Implications: This paper documents theoretical and other problems in PRIME along with CFD simulations and wind tunnel observations that support these findings. Although AERMOD/PRIME may provide accurate and unbiased estimates (within a factor of 2) for some building configurations, a major review and update is needed so that accurate estimates can be obtained for other building configurations where significant overpredictions or underpredictions are common due to downwash effects. This will ensure that regulatory evaluations subject to dispersion modeling requirements can be based on an accurate model. Thus, it is imperative that the downwash theory in PRIME is corrected to improve model performance and ensure that the model better represents reality.

Introduction

On April 21, 2000, the U.S. Environmental Protection Agency (EPA) proposed that AERMOD (American Meteorological Society/U.S. Environmental Protection Agency Regulatory Model) be adopted as a replacement to ISC3 (EPA, 1995, 2004). Based on public comments and peer review received at the Seventh Conference on Air Quality Modeling in June of 2000, EPA revised AERMOD to incorporate the Plume Rise Model Enhancements (PRIME) building downwash algorithms (Schulman et al., 2000) along with other changes not related to building downwash. In December of 2006, AERMOD (Cimorelli et al., 2005) officially became the EPA preferred model for regulatory dispersion modeling applications. The PRIME algorithm incorporates enhanced plume dispersion due to the turbulent wake behind sharp-edged rectangular buildings and reduced plume rise due to descending streamlines behind these obstacles and entrainment of the plume in the building cavity. PRIME calculates fields of turbulence intensity and wind speed, as well as the local slope of the mean streamlines as a function of the building dimensions, and, coupled with a numerical plume rise model, determines the change in plume centerline location with downwind distance.

Cimorelli et al. (2005) provides a brief description of PRIME and references Schulman et al. (2000), which is the only detailed documentation available on PRIME. No improvements to the downwash algorithms in PRIME have been made in more than 15 years since that original publication. Because building downwash often causes concentration predictions that exceed ambient standards, it is critical that these estimates be as accurate as possible. As discussed in the following paragraphs, recent field and wind tunnel studies have shown that AERMOD/PRIME can overpredict concentrations by factors of 2–8 for certain building types. On the other hand, for certain building and terrain configurations, AERMOD/PRIME can underpredict concentrations (Petersen and Beyer-Lout, 2012).

Several recent examples have documented that AERMOD tends to significantly overpredict maximum concentrations in some cases. Schulman and Scire (2012) showed that AERMOD-predicted maximum annual concentrations were at least 10 times greater than field observations for a very wide and long smelter in Tennessee. The hourly AERMOD-predicted maximum concentrations were 2–10 times greater than field observations. Schulman and Scire (2012) also ran AERMOD with and without a building for stacks well above the top of the building. This research found that for stack height to building height ratios (H_s/H_b) of 2.25–3.25 and a building length to height ratio (L/H_b) of 2.25, maximum concentrations with the building present were 3–14 times greater than those without the building present when building width to height ratios (W/H_b) range from 4 to 20. They also found that with a H_s/H_b ratio of 2.5, W/H_b ratio of 10, and L/H_b ratio of 3, maximum concentrations with the building present were about 3.5 time greater than those without the building. With an L/H_b ratio of 8.5, maximum concentrations were approximately 10 times greater than those without a building. These increases in maximum ground-level concentrations due to building effects are unreasonably high when compared with EPA’s (1985) wind tunnel testing results that showed concentrations increasing by 40–80% when the stack is equal to 2.5 times the building height. In addition, Huber (1989) conducted a series of wind tunnel tests and found that for H_s/H_b = 1.5, the maximum concentration increased by only a factor 2 for a building with W/H_b = 2. As W/H_b increased to 4 and 8, maximum ground-level concentrations decreased.

Baugues (2016) compared AERMOD predictions with observations at four monitoring stations near the Gibson power plant located in southwestern Indiana. Actual hourly sulfur dioxide (SO₂) emission rates for 2010 and meteorological data from an on-site meteorological tower were used as model input. Baugues concluded that AERMOD overpredicts by more than a factor of 2 for predictions over 35 parts per billion, where model performance is much more critical in evaluating compliance with the 1-hr SO₂ national ambient air quality standards (NAAQS). He also showed that in spite of using on-site meteorological data at three levels, when AERMOD predictions are paired in time and space with monitored concentrations, very poor agreement is shown (i.e., very low correlation coefficients). These results suggest that other factors, such as sub-hourly variations in meteorological conditions (i.e., wind speed and wind direction), source characterization, and inaccurate building downwash formulations, may be affecting model accuracy on a spatial and temporal basis.

Shea et al. (2007) reported on a field study where SO₂ concentrations were monitored on a residential tower located near the Mirant Power Station located in Alexandria, Virginia. Hourly average AERMOD-predicted concentrations were an order of magnitude greater than actual observations. Wind tunnel–determined equivalent building dimensions (EBDs) as described by Petersen and Reifschneider (2007) were used for this project. The EPA defines the EBD method as a source characterization technique (Tikvart, 1994; Bridgers, 2011) that can be used in place of the Building Profile Input Program (BPIP) to determine building input dimensions. In the case of the Mirant Power Station, the use of the EBD method resulted in better agreement between predicted values and field observations. However, the best agreement was found when building downwash effects were turned off (i.e., with no building present). This suggests that the theory in PRIME was assigning downwash effects to the stacks when in reality the stacks were unaffected by the building effects.

Oleson et al. (2009) compared AERMOD/PRIME and two other models against a wind tunnel data set for a simple stack-building configuration and found that AERMOD/PRIME showed a tendency to agree well or underpredict in the far field. In the near field, AERMOD/PRIME was shown to either overpredict or underpredict concentrations by a factor of more than 2. The highest overprediction tendency was observed for the short stack case with the cubic and long buildings.

De Melo et al. (2012) compared AERMOD/PRIME predictions with wind tunnel observations for a pig farm building complex. The pig farm barn complex consisted of two attached buildings 10.6 m and 7.56 m high with a combined width that ranged from 42 to 65 m depending on the wind direction. This research showed that, in general, AERMOD/PRIME agreed well with observations farther downwind except for one of the four wind directions evaluated where AERMOD overestimated maximum concentrations by about a factor of 2. In the near field, AERMOD overestimated by about a factor of 2 or more for one wind direction and underestimated observed concentrations for the other three wind directions evaluated. This study also found that AERMOD overpredicts maximum concentrations for specific building configurations at the plume centerline while underpredicting at the edges of the plume.

The discussion above suggests that there are theoretical and building characterization problems with PRIME and BPIP. Although a wind tunnel EBD study can be conducted to minimize the overprediction or underprediction bias, it is critical that PRIME be updated based on the latest research and understanding of building downwash effects so these source characterization techniques are not necessary to improve model performance. This paper documents theoretical and other problems that have been found in PRIME and provides computational fluid dynamics (CFD) simulations and wind tunnel observations that support these findings.

Importance of wake height calculation in PRIME

PRIME has equations for computing the wake height (H_w), cavity height (H_c), and plume rise (H_p). Figure 1 illustrates the results of these calculations for a simple building. If the plume rise at a given downwind distance is below H_w, then the PRIME building downwash algorithms are applied (i.e., enhanced turbulence, modified wind speeds, and modified plume trajectories). If the plume height is above H_w, then the standard no-building dispersion algorithms are applied. If a portion of the plume is below H_c, a cavity calculation is carried out. This evaluation found no significant problems with the cavity calculations.

Figure 1. Diagram showing the wake height, cavity height, and plume rise downwind of a simple building. The figure also illustrates the approach and wake mean velocity, vertical turbulence intensity, and vertical rms wind speed profile assumptions in PRIME for calculating turbulence intensity.

Display full size

Figure 1 shows a case where the plume from a stack is initially below H_w, causing the PRIME building wake algorithms to be applied. Farther downwind, the plume is above H_w and the no-building dispersion algorithms are in effect from that point on. This figure illustrates the importance of the H_w calculation along with how the turbulence and flow fields below H_w are computed.

Turbulence enhancement theoretical problems

As stated in Schulman et al. (2000), PRIME plume dispersion in the building wake region is based on a paper by Weil (1996). The vertical and horizontal turbulence intensities (i_z_o and i_y_o) of the flow approaching the building are enhanced to i_z and i_y starting at the lee wall of the building and decay back to i_z_o and i_y_o farther downwind based on a two-thirds power law variation with distance from the lee wall of the building. As discussed below, the turbulence intensity is unrealistically enhanced by a constant factor from ground level up to the height of the wake (H_w). The computed horizontal and vertical turbulence intensity values in the wake region (i_z and i_y) are then used in AERMOD to compute the horizontal and vertical dispersion coefficients (σ_y and σ_z) for ultimate use in computing the ground-level concentrations. In addition, PRIME assumes that the approach wind speed (U_o) has a maximum decrease, or deficit, starting at the lee wall and the deficit increases back to 1 (no deficit) farther downwind, also based on a two-thirds power law. For the purpose of calculating turbulence intensity, the velocity deficit factor is constant below H_w, whereas for calculating plume rise, the velocity deficit more realistically decreases with height above the building top. Figure 1 illustrates these concepts and shows some of the relevant variables.

Schulman et al. (2000) developed his equations for estimating vertical turbulence intensity (i_z) within the building wake region at a distance ξ from the lee wall based on the following initial equation: (1)

where i_z(ξ,y,z) is the vertical turbulence intensity in the building wake that starts at ξ/R = 1 and is below the wake boundary, ξ is defined such that ξ = R at the lee wall, R is a computed building length scale, σ_w(ξ,y,z) is the root-mean-square (rms) vertical velocity in the building wake region, U(ξ,y,z) is the mean wind speed in the building wake region, σ_wo(z) is the rms vertical velocity in the undisturbed flow, Δσ_wo is the difference between the local value at the lee wall of the building (i.e., at ξ = R) and the upstream value, U_o(z) is the mean velocity in the undisturbed flow, and ΔU_o is the velocity deficit (U − U_o) between value at the lee wall and the upstream value. It should be noted that the same relationships apply for the lateral turbulence intensity, i_y.

In the above definitions, many of the variables are functions of ξ, y, and z as indicated. As expected, in the building wake region, Δσ_wo and ΔU_o vary with ξ, y, and z, but PRIME assumes that they both vary with ξ with a step change function in the z direction at H_w (in the wake or out of the wake). There is a similar step change in the y direction when the plume is beyond the horizontal wake. It should be noted that PRIME does realistically assume ΔU_o varies with height above the building for the plume rise calculation.

Equation 1 can be further simplified to (2)

where i_z_o is the vertical turbulence intensity in the approach flow and ΔU_o/U_o = −0.7 and Δσ_wo/σ_wo = 0.7 based on Schulman et al. (2000). Based on eq 2, at the lee edge of the building, ξ = R and i_z = 5.7 i_z_o, and as ξ/R becomes large, i_z = i_z_o, or the vertical turbulence intensity returns to the undisturbed value.

Schulman et al. (2000), on the other hand, presents the following final equation for estimating turbulence intensity that was derived from eq 2: (3)

where Schulman et al. (2000) modified eq 2 by adding the term i_zN. In effect it is assumed, without supporting evidence, that 1.7 i_zN/i_z_o − 1 = Δσ_wo/σ_wo, which in turn only equals 0.7 when i_zN = i_z_o. When i_z₀ is less than i_zN, Δσ_wo/σ_wo is greater than 0.7 and the converse is true when i_z₀ is greater than i_zN. In addition, Schulman et al. (2000) assumes that i_zN (and i_yN) are defined by constant i_z_o and i_y_o values of 0.06 and 0.08, which are equivalent to vertical and horizontal standard deviations of wind angle values of 3.4 and 4.6 degrees. According to EPA (2000), these turbulence intensity values are representative for a surface roughness of 15 cm and Paquill-Gifford stability category E. The validity of modifying the Weil (1996) approach by adding the i_zN term should be evaluated, since proper scientific justification is lacking in Schulman et al. (2000). The impact of adding the i_zN term is that turbulence intensity at ξ = R (lee wall of the building) varies with stability as follows: i_z ~11 i_z_o for stable stratification, i_z ~ 5.7 i_z_o for neutral stratification and i_z ~ 3 i_z_o for unstable stratification. Although this may seem reasonable, that is, high turbulence enhancement occurs with a less turbulent approach flow, there is no basis or reference to confirm this assumption.

A more broadly applicable extension of Weil’s (1996) initial approach is provided in the equation below where Δσ_wo/σ_wo and ΔU_o/U_o vary with height, approach turbulence, stability, and structure type and/or shape. (4)

Some of the model performance problems noted above may be due to the over simplification of the turbulence intensity calculation method. Additional research is needed to determine how Δσ_wo/σ_wo and ΔU_o/U_o vary with height, approach turbulence intensity, stability, and structure type.

Depth of enhanced turbulence and velocity deficit region

PRIME assumes that the enhanced turbulence and velocity deficit region extends from ground level to the height of the wake (H_w). Figure 2 shows the turbulence enhancement region in PRIME versus the realistic enhancement region based on CFD simulations and wind tunnel measurements carried out by the authors. Past researchers (Peterka et al., 1985; Woo et al., 1977) have also shown that the turbulence enhancement (and velocity deficit) decreases with height above the top of the building.

Figure 2. Illustration of PRIME enhanced turbulence region (top) and realistic enhanced turbulence region (bottom) for a building with height to length to width ratio of 1:1:8. The top panel shows that PRIME assumes that the turbulence is enhanced by a constant factor starting at the lee edge of the building and decreases with downwind distance. The bottom panel shows a more realistic illustration where the turbulence is enhanced starting at the lee edge of the building up to the height of the building and then quickly decays back to ambient turbulence levels. Also, the turbulence is enhanced above the roof of the building in the bottom panel where PRIME assumes no enhancement in this region.

Display full size

Figure 3 shows velocity contours averaged over 300 sec of simulated time from a CFD simulation using the Fire Dynamics Simulator (McGratten et al., 2013) run in large eddy simulation (LES) mode for a 15-m-high (H_b) building with H_b to W to L ratio of 1:1:2. The model domain is 120 m along wind by 100 m crosswind by 72 m high; the resolution (grid cell size) is 0.3 m by 1.2 m by 0.5 m (2.25 million cells); and the building dimensions are 15 m high by 30 m long by 15 m wide. The boundary conditions and other assumption are as follows:

Inlet wind profile: U_o= 6.8 (z/72 m)^0.272 (m/sec)
Outlet: ambient pressure boundary condition
Top: tangential velocity boundary matching wind profile
Sides: solid, inert
Large eddy simulation (transient)
Deardorff Sub-grid Turbulence Model
Superbee TVD flux limiter

Figure 3. FDS LES simulation of airflow around a 15-m-high rectangular structure with building aspect ratio (H_b:W:L) of 1:1:2. The figure shows average velocity contours where darker areas indicate a slower wind speed. The flow is from the left to the right. The darker areas to the right of the building show the region of low wind speed or velocity deficit. The figure scale can be referenced to the building, which is 15 m high and 30 m long.

Display full size

Figure 3 shows that the region of low velocity when compared with the upstream velocity (i.e., velocity deficit) only extends slightly above the top of the building versus extending up to the height of the wake, which would be well above the top of the building as illustrated in Figure 2.

Figure 4 shows the vertical turbulence intensity enhancement based on velocity measurements obtained in a boundary-layer wind tunnel for the same 1:1:2 (i.e., H_b to W to L ratio) building. These measurements were carried out by the authors in the CPP wind tunnel laboratory. The approach wind profile was representative of a site with a surface roughness of approximately 0.30 m, full scale. The 15-m-high building was simulated at a 1:50 model scale. Figure 4 shows that the turbulence intensity enhancement only extends slightly above the top of the building and decays rapidly back to ambient turbulence levels. The figure shows the maximum i_z/i_z_o ~ 5.5 at ξ/R = 1.5, with an average value of approximately 4 below the building top (z/H_b =1), decreasing rapidly to ambient (i.e., 1) above the building top. PRIME and eq 2 would give an i_z/i_z_o value of 4.4, which agrees well with the average value below the building top in Figure 2, but PRIME would apply this same value up to the wake height (H_w), which would be in the range of z/H_b equal to 1.5–2.0 in the figure. This means that for many structures, the depth of the high turbulence region in PRIME can be well above the building roof as shown in Figure 2. This means that taller stacks and/or higher plume rise are required to escape the exaggerated and unrealistic building downwash region in the model. This may explain the overprediction problems noted by Schulman and Scire (2012) and Petersen and Beyer-Lout (2009, 2012) for long and/or wide structures.

Figure 4. Vertical turbulence intensity enhancement factor (i_z/i_z_o) versus normalized height (z/H_b) at four normalized distances (ξ/R = 1, 1.5, 2, and 3) based on wind tunnel measurements conducted by the authors for a simulated 15-m-high (H_b) building that is 15 m wide and 30 m long. i_z is the vertical turbulence intensity at height z/H_b in the lee of the building, and i_z_o is the vertical turbulence intensity at the same height in the approach flow. Note that the maximum i_z/i_z_o ~ 5.5 at ξ/R = 1.5 which agrees well with the eq 2 prediction of 5.7 at ξ/R = 1.0. PRIME assumes a constant i_z/i_z_o value of 5.7 at ξ/R = 1.0 from ground level up to wake height (i.e., at z/H_b 1.5 to 2.0 in the figure).

Display full size

Other problems

Building dimension inputs

The BPIP preprocessor (EPA, 1993) frequently creates artificially large buildings, as illustrated in Figure 5. Since the formulation in BPIP creates an artificially large building when wind blows at an angle, the starting point for the wake growth moves farther upwind (location A versus location B in Figure 5). This means that the height of the wake is much higher at the lee edge of the building than it should be if the wake growth started at location B. In addition, the building wake turbulence enhancement should in reality start at location C, whereas PRIME will assume it starts at location D. This results in an overstated wake height at location D and an overstated amount of turbulence enhancement. The overstated wake height is illustrated in Figure 5 at locations C and D by the lines noted “AERMOD Wake Boundary” and “Realistic Wake Boundary.” Regarding the effect on turbulence intensity, the approach turbulence intensity is increased by a factor of 5.7 at locations C from ground level up to the realistic wake height based on eq 2. Using BPIP building input dimensions, the turbulence intensity would not be enhanced until the plume reaches Location D where the approach turbulence would be enhanced by a 5.7 factor. In reality, the approach turbulence should only be enhanced by a factor of 2.6 at location D (ξ/R ~ 2) based on eq 2.

Figure 5. Plan view (top) of an actual building (dark color) and the BPIP generated building (gray) when wind flows at an angle to the building and corresponding elevation view (bottom) showing the resulting AERMOD/PRIME and realistic wake heights and enhanced turbulence zones.

Display full size

PRIME building dimension inputs could be improved by updating the BPIP program to compute more realistic building dimension inputs using a method such as that discussed in Lefebvre et al. (2013), which is based on the method used in the Danish Operationelle Meteorologiske Luftkvalitetsmodeller (OML) (Olesen and Genikhovich, 2000). This method determines the building length (L) by drawing a line aligned with the wind direction and intersecting the source. The length of the portion of that line that is within the building is taken to be L. In Figure 5, L would be the length between points B and C. The building width is then taken to be the maximum length of a line perpendicular to the straight line used to define L that is completely within the building. In Figure 5, W would be the length of the line through the stack that is included within the building that is perpendicular to the line through points B and C. In the case shown in Figure 5, the building width would be much smaller than the projected maximum width illustrated in the figure and would not characterize the horizontal wake correctly. A better approach could be to use the Olesen and Genikhovich (2000) method to determine L but then use the maximum projected width or a width that preserves the initial building volume. Additional research is needed to evaluate the best approach.

Complex buildings

For stacks located on or near several buildings of different heights and shapes, BPIP has analytical and logical relationships that merge buildings and ultimately calculate one effective width, length, height, and position to represent all buildings for each of 36 wind directions relative to the stack. Clearly, the PRIME algorithm cannot replicate these complex environments as noted by de Melo et al. (2012) and Petersen and Beyer-Lout (2012). For these situations, more advanced modeling such as computational fluid dynamics or wind tunnel modeling is needed.

Streamline slope discontinuity

Schulman et al. (2000) provides the following two equations for computing the streamline slope in two regions of the flow. (5) (6)

At x = 0.0, these two equation should provide the same result but they are different by a factor of 2 as illustrated below. (7) (8)

It is not clear what effect this will have on concentration predictions; however, this discontinuity may need to be corrected. It is likely that the constants 2 and 4 at the front of the equations should be equal, but it is not clear which is the correct value. This will likely only affect concentration predictions for stacks that are slightly taller or shorter than the building and located upwind of the leading edge of the building.

Streamline calculation for lattice and streamlined structures

Lattice/porous and streamlined structures are also not modeled correctly in AERMOD/PRIME, since the model assumed that these structures are rectangular solids. This means that the flow and turbulence fields downwind of the structures will not be properly characterized. Petersen and Beyer-Lout (2009) demonstrated that plume dispersion is enhanced with lattice structures upwind or downwind from a stack, but in general the plume travels in a horizontal direction. This research also showed that when a porous structure is replaced with a similarly sized solid structure placed upwind or downwind from a stack, the plume dispersion is much greater. This results in higher predicted maximum concentrations because the plume hits the ground much closer to the stack.

Petersen et al. (2014) conducted wind tunnel and AERMOD modeling for a single cylindrical storage tank. That study showed that the AERMOD predictions based on the current treatment of cylindrical structures (i.e., as rectangular solids) are lower than those when the cylindrical structure is properly modeled in AERMOD using a wind tunnel–determined EBD. This may be due to the fact that the horizontal wake is smaller for a cylinder than for a rectangular solid of the same width. This suggests that the lower ground-level concentrations predicted for the rectangular solid may be due to greater horizontal dispersion than occurs for a cylindrical structure.

To better handle these types of structures, the streamlines generated by PRIME could be made horizontal for lattice or porous structures, whereas for cylindrical structures more research is needed to define the streamline characteristics. Additional research is also needed to define appropriate Δσ_wo/σ_wo and ΔU_o/U_o values for these types of structures.

The corner vortex

The BPIP program translates structures for any angle of approach into a single rectangular block using the projected width of the structure with the leading wall oriented perpendicular to the wind direction. There is no distinction made for different angles of approach once the rectangular block is calculated; hence, the current building wake equations do not account for the case when the wind blows along a building corner. In this instance, two vortices are generated that counterrotate such that there is a downward velocity generated downwind of the building. This downward velocity will tend to decrease plume rise downwind of the building, thereby resulting in increased ground-level concentrations. This has been documented in Petersen et al. (2012) where a detailed discussion on this topic is provided, including an example where a field monitor exhibited maximum concentrations 2 times higher than AERMOD predictions. The difference was attributed to the corner vortex effect, and results from wind tunnel simulations confirmed the same AERMOD underprediction tendency.

Upwind terrain wakes

Even though AERMOD incorporates terrain, it is only used to characterize the direct interaction of the plume with terrain features downwind of the stack. The wakes and eddies generated by terrain upwind of the stack are not considered. Since AERMOD allows for the placement of a building some distance upwind of the stack, wind tunnel-determined EBDs can be used to account for the upwind terrain wake effect. Petersen et al. (2007) describes such a study where EBD values were determined to account for upwind terrain. When the EBD values were input into AERMOD, the overall maximum concentration increased by nearly a factor of 2 compared with the case without the EBD values to account for the upwind terrain. Until the theory is improved to account for upwind terrain wake effects, wind tunnel-determined EBD values can be used along with terrain amplification factors applied to the AERMOD predictions as discussed in Petersen and Beyer-Lout (2012).

Summary and conclusions

This paper documents theoretical and other problems in PRIME along with CFD and wind tunnel simulations that support these findings. Although PRIME may provide unbiased estimates (within a factor of 2) for some building configurations, it is critical that it be updated based on the latest research and understanding of building downwash effects to accurately estimate concentrations for building configurations where significant overpredictions or underpredictions due to downwash effects are common. This will ensure that regulatory evaluations subject to dispersion modeling requirements can be based on an accurate model. Thus, it is imperative that the downwash theory in PRIME be corrected to improve model performance for a wider variety of building types.

A plan to address some of the downwash problems presented herein is currently in progress through the PRIME2 Advisory Subcommittee formed by the Atmospheric Modeling and Meteorology (APM) subcommittee from the Air and Waste Management Association (A&WMA). This effort includes the collaboration of technical experts, industry groups, and representatives from the EPA. The two main objectives of this effort include (1) providing a technical review forum to improve the PRIME building downwash algorithms; and (2) establishing a mechanism to review, approve, and implement new science into the model. The scope of the first phase of this project includes updating the formulation in PRIME to more accurately characterize porous and elongated buildings, correcting known theoretical issues, and correcting bugs identified through a thorough review of the existing code.

Related Research Data

Wake Characteristics of Tall Buildings in a Realistic Urban Canopy

Source: Springer Science and Business Media LLC

Linking provided by

Additional information

Notes on contributors

Ron L. Petersen

Ron L. Petersen, Sergio A. Guerra, and Anthony S. Bova are engineering and environmental consultants specialized in atmospheric dispersion modeling at CPP, Inc. Fort Collins, CO, USA.

Sergio A. Guerra

Ron L. Petersen, Sergio A. Guerra, and Anthony S. Bova are engineering and environmental consultants specialized in atmospheric dispersion modeling at CPP, Inc. Fort Collins, CO, USA.

Anthony S. Bova

Ron L. Petersen, Sergio A. Guerra, and Anthony S. Bova are engineering and environmental consultants specialized in atmospheric dispersion modeling at CPP, Inc. Fort Collins, CO, USA.

References

Baugues, K. 2106. An assessment of the American Meteorological Society/U.S. EPA Regulatory Model’s (AERMOD’s) accuracy: A case study—revised. Indianapolis, IN: Indiana Department of Environmental Management Office of Air Quality. January 2016. https://www.in.gov/idem/airquality/files/aermod_assessment_report_2016.pdf (accessed February 6, 2017). [Google Scholar]
Bridgers, G. 2011. Model clearinghouse review of equivalent building dimensions for AERMOD. Research Triangle Park, NC: U.S. Environmental Protection Agency. October 24, 2011. https://www3.epa.gov/ttn/scram/guidance/mch/new_mch/11-VII-01_MCResponse_Region7_EBDs_10242011.pdf (accessed February 6, 2017). [Google Scholar]
Cimorelli, A.J., S.G. Perry, A. Venkatram, J.C. Weil, R.J. Paine, R.B. Wilson, R.F. Lee, W.D. Peters, and R.W. Brode. 2005. AERMOD: A dispersion model for industrial source applications. Part I: General model formulation and boundary layer characterization. J. Appl. Meteorol. 44:682–93. doi:10.1175/JAM2227.1 [Crossref], [Google Scholar]
de Melo, A.M.V, J.M. Santos, I. Mavroidis, and N.C. Reis Jr. 2012. Modelling of odour dispersion around a pig farm building complex using AERMOD and CALPUFF. Comparison with wind tunnel results. Build. Environ. 56:8–20. doi:10.1016/j.buildenv.2012.02.017 [Crossref], [Web of Science ®], [Google Scholar]
Huber, A.1989. The influence of building width and orientation on plume dispersion in the wake of a building. Atmos. Environ. 23:2109–16. doi:10.1016/0004-6981(89)90172-8 [Crossref], [Web of Science ®], [Google Scholar]
Lefebvre, W., G. Cosemans, and J. Kegels. 2013. Comparison of the IFDM building downwash model predictions with field data. Atmos. Environ. 75:32–42. doi:10.1016/j.atmosenv.2013.04.022 [Crossref], [Web of Science ®], [Google Scholar]
McGratten, K., R. McCermott, and C. Weinschenk. 2013. Fire Dynamics Simulator Technical Reference Guide, Volume 1: Mathematical Model. National Institute of Standards and Technology Special Publication 1018-1. October 2013. Gaithersburg, MD: National Institute of Standards and Technology. [Crossref], [Google Scholar]
Olesen, H.R., R. Berkowicz, M. Ketzel, and P. Løfstrøm. 2009. Validation of OML, AERMOD/PRIME and MISKAM using the Thompson wind-tunnel dataset for simple stack-building configurations. Bound. Layer Meteorol. 131:73–83. doi:10.1007/s10546-009-9355-9 [Crossref], [Web of Science ®], [Google Scholar]
Olesen, H.R., and E. Genikhovich. 2000. Building Downwash Algorithm for the OML Atmospheric Dispersion Model, 30. Research Notes from NERI No. 123. Denmark: National Environmental Research Institute. http://www2.dmu.dk/1_viden/2_Publikationer/3_arbrapporter/rapporter/AR123.pdf (accessed February 6, 2017). [Google Scholar]
Peterka, J.A., R.N. Meroney, and K.M. Kothari. 1985. Wind flow patterns about buildings. J. Wind Eng. Ind. Aerodyn. 21:21–38. doi:10.1016/0167-6105(85)90031-5 [Crossref], [Web of Science ®], [Google Scholar]
Petersen, R.L., and A. Beyer-Lout. 2009. Is AERMOD/PRIME overpredicting for short buildings with a large footprint? Technical Paper 517. Presented at the 102nd Annual A&WMA Conference & Exhibition, Detroit, MI, June 16–19, 2009. http://www.cppwind.com/wp-content/uploads/2014/01/AWMA-2009-Petersen-Paper-517.-Large-Foot-Print.pdf (accessed February 6, 2017). [Google Scholar]
Petersen, R.L., A. Beyer-Lout, R. J. Paine. 2012. Evaluation of monitored SO₂ NAAQS exceedances due to the corner vortex. Technical Paper 451. Presented at the 105th Annual A&WMA Conference & Exhibition, San Antonio, TX, June 19–22, 2012. http://www.cppwind.com/wp-content/uploads/2014/01/AWMA-2012-Petersen.Paper-451.Monitored-Exceedance.pdf (accessed February 6, 2017). [Google Scholar]
Petersen, R.L., A. Beyer-Lout. 2012. AERMOD building downwash theoretical limitations and possible solutions. Technical Paper 387. Presented at the 105th Annual A&WMA Conference & Exhibition, San Antonio, TX, June 19–22, 2012. http://www.cppwind.com/wp-content/uploads/2014/01/AWMA2012-Petersen-Paper-387-Downwash-Problems.pdf (accessed February 6, 2017). [Google Scholar]
Petersen, R.L., A. Kolesnikov, and A. Beyer-Lout. 2014. Evaluation of new wind tunnel, AERMOD and CFD methodologies to determine EBD/BPIPPRM in light of the 2011 EPA Memorandum. Technical Paper 33263. Presented at the 107th Annual A&WMA Conference & Exhibition, Long Beach, CA, June 24–27. https://www.researchgate.net/publication/286404722_Evaluation_of_New_Wind_Tunnel_AERMOD_and_CFD_Methodologies_to_Determine_EBDBPIP_in_Light_of_the_2011_EPA_Memorandum (accessed February 6, 2017). [Google Scholar]
Petersen, R.L., and J. Reifschneider. 2007. Improved building dimension inputs into AERMOD modeling of the Mirant Potomac River Generating Station. Technical Paper 276. Presented at the 100th Annual A&WMA Conference & Exhibition, Pittsburgh, PA, June 2007. https://www.researchgate.net/publication/255661205_Improved_Building_Dimension_Inputs_for_AERMOD_Modeling_of_the_Mirant_Potomac_River_Generating_Station (accessed February 6, 2017). [Google Scholar]
Petersen, R.L., J. Reifschneider, R.J. Paine, and K. Schmidt. 2007. Use of equivalent building dimensions to characterize upwind terrain wake effect for AERMOD. Technical Paper 425. Presented at the 100th Annual A&WMA Conference & Exhibition, Pittsburgh, PA, June 2007. http://www.cppwind.com/wp-content/uploads/2014/03/EBD_TerrainWakes_AWMA_07.pdf (accessed February 6, 2017). [Google Scholar]
Schulman, L.L., D. Strimaitis, and J.S. Scire. 2000. Development and evaluation of the PRIME plume rise and building downwash model. J. Air Waste Manage. Assoc. 50:378–90. doi:10.1080/10473289.2000.10464017 https://www3.epa.gov/scram001/7thconf/iscprime/tekpapr1.pdf (accessed February 6, 2017). [Taylor & Francis Online], [Web of Science ®], [Google Scholar]
Schulman, L.L, and J.S. Scire. 2012. Building downwash modeling with AERMOD. Public comments. Presented at the 10th Conference on Air Quality Modeling, EPA Research Triangle Park, NC, March 3–5, 2012. https://www3.epa.gov/scram001/10thmodconf/presentations/3-3-EPA10thConferenceAQModeling-BuildingDownwash_SchulmanScire.pdf (accessed February 6, 2017). [Google Scholar]
Shea, D., O. Kostrova, A. MacNutt, R. Paine, D. Cramer, L. Labrie. 2007. A model evaluation study of AERMOD using wind tunnel and ambient measurements at elevated locations. Technical Paper 198. Presented at the 100th Annual A&WMA Conference, Pittsburgh, PA, June 26–29, 2007. [Google Scholar]
Tikvart, J.A. 1994. Chief, Source Receptor Analysis Branch, U.S. Environmental Protection Agency, letter to Brenda Johnson, Regional Modeling Contact, Region IV and Douglas Neeley, Chief Air Programs Branch, Region IV, July 25, 1994. https://www3.epa.gov/scram001/guidance/mch/cfym101.txt (accessed February 6, 2017). [Google Scholar]
U.S. Environmental Protection Agency. 1985. Guideline for determination of good engineering practice stack height (technical support document for the stack height regulations). EPA-450/4-80-023R. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards. [Google Scholar]
U.S. Environmental Protection Agency. 1993. User’s guide to the building profile input program. EPA-454/R-93-038. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards, Technical Support Division. [Google Scholar]
U.S. Environmental Protection Agency. 1995. User’s guide for the Industrial Source Complex (ISC3) dispersion model. EPA-454/B-95-003a. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards. [Google Scholar]
U.S. Environmental Protection Agency. 2000. Meteorological monitoring guidance for regulatory modeling applications. EPA-454/R-99-005. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards. [Google Scholar]
U.S. Environmental Protection Agency. 2004. AERMOD: Description of model formulation. EPA-454/R-03-004. Research Triangle Park, NC: U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards. [Google Scholar]
Weil, J.C. 1996. A new dispersion model for stack sources in building wakes. Technical paper presented at the Ninth Joint Conference on Applications of Air Pollution Meteorology with the Air & Waste Management Association, Atlanta, GA, January 28–February 2, 1996. [Google Scholar]
Woo, H.G.C., J.A. Peterka, and J.E. Cermak. 1977. Wind tunnel measurements in the wakes of structures. NASA Contractor Report NASA CR 2806. https://ntrs.nasa.gov/archive/nasa/casi.ntrs.nasa.gov/19770012772.pdf (accessed February 6, 2017). [Google Scholar]

Journal of the Air & Waste Management Association

Critical review of the building downwash algorithms in AERMOD

Technical Papers

Critical review of the building downwash algorithms in AERMOD

ABSTRACT

Introduction

Importance of wake height calculation in PRIME

Published online:

Turbulence enhancement theoretical problems

Depth of enhanced turbulence and velocity deficit region

Published online:

Published online:

Published online: