BioWin Advantage 10.1 : Investigating Chemical Phosphorus Removal in BioWin 6

Introduction

In this issue we will be exploring modeling of chemical phosphorus (P) removal in BioWin 6. BioWin 6 incorporates a new model for chemical P removal with metal salts addition. Detailed information about the new chemical P removal model is found in Appendix A. We will look at the basic requirements and important checks for setting up a configuration with metal salts addition for P removal. We will then look at a configuration to investigate the efficiency of chemical P removal.

New Chemical P Removal Model in Brief

The chemical P removal model mechanisms are as follows:

• Addition of metal results in the instantaneous formation of hydrated metal oxides that have a high number of active surface sites (HMO-High) for interaction with P.
• If P is present it can co-precipitate with the HMO-High. At the same time, the floc that is formed begins to change; it begins to lose potential P adsorption sites as it ages. This “low site” floc is tracked in the model as a separate species (HMO-Low).
• P adsorbs to the remaining floc with a lower available number of sites with time. P can adsorb to HMO-Low, but that adsorption rate is slower than the co-precipitation rate for P interaction with HMO-high. Having the majority of P interact with high surface species will optimize P removal.
• The low site species also simultaneously begin to change; they transform to an “aged” species which cannot take up P.
• Aging is inversely proportional to mixing intensity. The lower the mixing intensity, the faster the aging process occurs.
• In summary, two processes take away available sites: (1) they are “filled up” with adsorbed P, and (2) they “age out”.

These processes mean that Solids retention time (SRT) is better accounted for with the new model. Since the reactions are not instantaneous equilibria, so there is interaction between metal species and P as the high and low surface metal oxides flow around the system – even after metal dosing is turned off!

Metal-P interactions are also influenced by other factors as well:

1. pH

• pH dictates the species of P available for interaction (only H2PO4- is removed in the model).
• When ferric is added, H+ competes with H2PO4- for active surface sites at low pH.

2. Residual P concentrations

• The rates of co-precipitation and adsorption decrease as the residual P decreases. This helps the model to account for the influence of dose location on P removal performance. As a result, adding metal at the upstream end of a wastewater treatment process train is more efficient than adding metal at the downstream end. Multiple dosing points are better modeled with different removals in P from point to point.

3. Mixing

• Mixing intensity at the dosing point and throughout the process impacts model results.

More details about the new chemical P removal model can be found in Appendix A.

Setting Up Metal Addition – Important Checks

Important things to remember when using the new chemical P removal model in BioWin 6:

#1 Ensure metal is added to an element with volume

For the reactions described above to take place, metal must be added into an element that has volume. If metal is added into an element that does not have volume, (e.g. to a mixer or point clarifier element), soluble metal will flow through the element unreacted until reaching an element with volume. A new channel element is available to mimic metal addition into channels. More information on the channel element can be found in Appendix B.

#2 Check that the metal dose makes sense

Consult the literature for typical metal doses. Various rules of thumb and dosing charts are quoted in the industry or available in the literature.

For example, a molar ratio of 1.6 mol Fe/mol P and 0.8 mol Al/mol P was suggested in the work of Jenkins – Hermanowicz. This corresponds to a mass ratio of (1.6 mol Fe x 55.84 g Fe/mol) / (1 mol P x 30.97 g P/mol) = 2.88 kg Fe/kg P or (0.8 mol Al x 26.98 g Al/mol) / (1 mol P x 30.97 g P/mol) = 0.7 kg Al/kg P.

Another example practical guideline is that 20 kg of 40% FeCl3 solution is required to remove 1 kg of P. The FeCl3 content of 20 kg of the 40% solution would be 20 kg x 0.4 = 8 kg FeCl3. The Fe content is 8 kg FeCl3 x 55.84 mol Fe / 162.19 mol FeCl3 = 2.75 kg Fe. That is, a mass ratio of 2.75 kg Fe/kg P removed.

It is important to keep in mind that most of the literature values are based on jar testing with a goal of achieving effluent P of about 1 mg/L; achieving lower effluent TP typically requires a higher dose ratio. In addition there are many competing reactions that come into play (i.e. with colloids, solids, organics and/or humic substances). Therefore bench scale testing is often recommended to confirm dose requirements. Metal doses predicted when targeting very low residual P concentrations should be interpreted / used with a degree of caution. If both metal dose amount and effluent P concentration data are available, and model predictions to do not align with expected data, consider adjusting the high and low active site factors found in Project > Parameters > Physical/Chemical on the Fe constants or Al constants tab. If dosing occurs in very short retention time zones, consideration also may be given to adjusting the coprecipitation and adsorption rates found in Project > Parameters > Physical/Chemical on the Fe rates or Al rates tab.

When characterizing metal dose amount it is important to separate out the level of P removed via metal dosing versus the level of P removed via biomass synthesis uptake. A good check is to turn off metal dosing and determine how much P is removed without metal addition. Then determine the amount of P removal with metal addition. The difference in P removal without metal addition and with metal addition is the actual P removed by metal. This will be explored in the example below.

#3 Check the pH

Keep an eye on the pH to ensure an optimal range for both activated sludge and P removal is maintained. H2PO4- is the only P species that gets removed in the model. Based on the speciation of phosphoric acid with pH (see figure below), as the pH increases above 6 the availability of H2PO4- in solution decreases. When ferric is added, as the pH falls closer to 6, H+ begins to compete with H2PO4- for active sites thus lowering the efficiency for P removal.

#4 Ensure mixing is specified in all basins

Adjusting the mixing power can help to increase or decrease the level of P removed. Mixing can be provided in the model in three ways: (1) by flow, (2) by aeration, (3) by specifying a mixing power in bioreactor elements OR a velocity gradient in channel elements (see Appendix B). Hovering your cursor over a bioreactor/channel element will show power dissipation via mixing in the bottom right summary pane (see screen capture below). The power dissipation via mechanical mixing and air flow is also shown along with the velocity gradient in the unit.

When adding metal into a basin that does not have mixing, i.e. settler or solid/liquid separation units, consider adding a channel or bioreactor upstream of the unit to represent the mixing zone in the unit.

#5 Check metal species state variables

To help troubleshoot files with metal addition, look at the state variables for the hydrated metal oxide species. Metal species can be tabulated and/or charted in the BioWin Album, or viewed in BioWin’s Explorer. A list of the relevant metal state variables is provided below. Some tips on what to look out for include:

• The presence of unbound HMO (i.e. non-aged hydrated metal oxides without H2PO4- or H+ adsorbed) indicates some additional potential for P uptake.
• If no unbound hydrated oxides are available, and further P uptake is required, then consider increasing the dose of metal in the system.
• When dosing ferric, if you notice hydrated ferric oxides with H+ adsorbed consider increasing the pH to eliminate H+ competition for P removal.
• The examples below will investigate the hydrated metal oxide state variable distribution in a chemical P removal system.

#6 What about settlers?

In addition to considering the use of a channel or bioreactor element upstream of a settler element with volume, you may want to consider the impact of turning on/off reactions via the Model tab in the model or ideal settling unit. With chemical P removal, turning on reactions in an ideal or model clarifier may allow for further metal-P interactions. The extent of reactions is dependent on the type of setter element used. For example, in ideal clarifier units, a percentage of the metal hydroxides will be separated out of solution instantaneously as a function of the % solids capture specified. The extent of reactions via metal hydroxides remaining in the liquid phase will typically be low, especially if the sludge layer of the ideal settler is small. In a model clarifier unit, the sludge layer gets divided into a number of layers and reactions are more plug flow as solids settle from one layer into the other. This provides more opportunities for metal-P interactions compared to the ideal clarifier element. Note, depending on the configuration and SRT, turning on reactions could result in additional biomass decay causing P concentration to increase instead of decrease; as such turning on reactions is not always a sure way to further increase P removal.

#7 Channel Considerations

If using a channel element, make sure the linear velocity and HRT are sufficient. The dimensions of the channel should be modified to provide a residence time of at least 1 minute and a linear velocity of at least 0.3 m/s or 1 foot/s (i.e. often used in practice as a minimum mixing to keep MLSS in suspension). See Appendix B for more information on setting up Channels in BioWin.

Investigating Chemical Phosphorus Removal with Ferric

The BioWin configuration shown below has an influent flow of 24,000 m3/d, an influent TP concentration of 6.5 mg P/L, an influent pH of 7.3, and an SRT of 11 days. A ferric element is used to dose 180 kg Fe/d into an influent channel. The influent channel is set up to provide a linear velocity of approximately 0.56 m/s and a residence time of approximately 1.2 minutes. Within the Operating tab of the channel’s property dialog box, the mixing at the point of metal addition is specified with a velocity gradient of 150/s. The mixing in the remaining zones is the velocity gradient associated with the power dissipated by the liquid flow (~115/s). Reactions are not active in the model clarifier.

Running the steady state results in an effluent pH of 6.84 and an effluent soluble PO4-P concentration of 0.43 mgP/L. The effluent and WAS stream PO4-P mass rates are 9.93 + 0.39 = 10.32 kgP/d. In order to understand how much P is removed via metal dose versus synthesis uptake the simulation was also run without metal addition. Setting the ferric addition flow to zero results in an effluent soluble PO4-P concentration of 3.10 mgP/L, and effluent and WAS stream PO4-P mass rates of 71.71 + 2.79 = 74.5 kgP/d. The difference in soluble PO4-P mass rate in the output stream with and without metal dosing is 74.5 – 10.32 = 64.18 kgP/d. So the addition of 180 kgFe/d corresponds to a mass ratio of 180/64.18 = 2.8 kgFe/kgP removed, or a molar ratio of (180/55.845)/(64.18/30.97) = 1.56 mol Fe/mol P removed. The dose amount appears to be in range of typical values quoted in literature.

The resulting TP and soluble P profiles are illustrated below. Approximately 5.8 mg P/L of TP (2.9 mg P/L of soluble P) are removed. The majority of the removal occurs at the point of metal addition i.e. within the influent channel as a result of co-precipitation with HFO-High surface, while a smaller amount is removed downstream as a result of adsorption onto HFO-Low surface.

The distribution of unbound and bound HFO is illustrated in the charts below. In the influent channel, a small amount of residual unbound HFO-High surface remains but is used up downstream. HFO-Low surface and HFO-Aged make up the majority of the unbound species found in the mixed liquor. In the event that a higher iron dose is not added to the system, or if iron dosing stops completely, this residual unbound HFO-Low surface will provide additional adsorption capacity until it is used up through P adsorption/aging and effluent soluble P breakthrough occurs. The amount of residual unbound HFO-Low surface depends on the mixing and residence time which will dictate the level of aging that occurs. The majority of the bound P is associated with HFO-High surface since the majority of removal occurs via co-precipitation. Since the pH in this process are close to neutral, interactions between HFO and H+ are not observed.

The presence of unbound HFO indicates that there is some residual adsorption capacity in the system. To test this we can turn off iron dosing and determine how long it takes for P to breakthrough in the effluent. The figure below demonstrates the dynamic case where iron addition to the process is stopped, and the effluent soluble P is monitored with time after the iron dose cessation. To generate this dynamic response, the following steps were taken:

1. A steady state simulation was run with ferric dosing on.
2. A dynamic simulation was run for 5 days with ferric dosing on. In the chart below, the dynamic simulation starts on 06/07/18.
3. Ferric dosing was turned off, and the simulation was continued. In the chart below, the ferric dose is stopped on 06/12/18.

The effluent soluble P concentration is shown below. The soluble P concentration remains low for a few days after metal addition stops on 06/12/18, and then begins increasing on 06/16/18 as the unbound metal hydroxide that persists in the system is used up through P adsorption/aging. The amount of P removal observed or the time that it takes for P breakthrough to occur after cessation of metal addition can be calibrated by adjusting the high and low active site factors (i.e. ferric active site factor (high) and ferric active site factor (low) found in Project > Parameters > Physical/Chemical > Fe Constants).

Optimizing Chemical Phosphorus Removal

Model versus Ideal Clarifier with and without Reactions

The example above was simulated without reactions on in the model clarifier. The table below explores simulation results with and without reactions on in both an ideal and model clarifier for comparison. Note the ideal clarifier case looks at the results of replacing the model clarifier in the configuration above with an ideal clarifier element.

From the results summarized in the table below it is evident that turning on reactions helps to lower the effluent soluble P concentrations. As discussed previously, when reactions are on further reactions occur, resulting in less unbound HFO with high and low sites remaining in the system and hence an increase in the amount of aged HFO. In this example, turning on reactions helps to encourage further metal-P interactions. However as mentioned above, this is not always the case - especially in long SRT systems or systems with biological P removal where further interactions may result in P release. Therefore, it is important to explore this on a case by case basis.

Impact of pH

Taking the original example file which used a model clarifier without reactions, the impact of pH on chemical P removal was tested by modifying the influent alkalinity concentration. The results of testing different influent alkalinity concentrations are summarized in the table below. Doubling the influent alkalinity from a value of 6 mmol/L to a value of 12 mmol/L resulted in an effluent pH of 7.27 compared to 6.88 in the base case. Increasing the pH caused less P to be removed (i.e. effluent soluble P of 0.63 mg P/L versus 0.43 mg P/L in the base case) since less H2PO4- is available in solution as pH increases above 6. As a result of having less P removed a lower concentration of HFO species with H2PO4-, and a higher concentration of unbound HFO species is observed in comparison to the base case. Reducing the influent alkalinity to 4 mmol/L resulted in an effluent pH of 6.46 and higher P removal (i.e. effluent soluble P of 0.35 mg P/L versus 0.43 mg P/L in the base case). Since more P was removed a higher concentration of HFO species with H2PO4- and a lower concentration of unbound HFO species is observed compared to the base case.

Impact of Mixing
In the base case example, a velocity gradient of 150/s is specified in the first zone of the channel element and a mixing power of 5 W/m3 is specified in the anoxic zone. If mixing was not specified in the channel element or power tab of the anoxic zone, mixing would still be provided in these elements via liquid flow. With mixing in the channel and in the anoxic zone set to zero, the velocity gradient due to the flow of liquid is 115.63/s in the channel and 38.36/s in the anoxic zone. In the aeration basins the velocity gradient results from both the flow and aeration; the gradient is 138.45/s in the first aerobic zone and 106.44/s in the second aerobic zone. Without any additional mixing specified in the channel element and anoxic zone the effluent PO4-P increases very slightly from 0.43 mg P/L to 0.44 mg P/L. This very small change indicates that the velocity gradient provided via the flow of liquid in this example is high enough to limit the extent of aging and its consequent reduction of P removal sites.If additional mixing is specified in the channel element and all of the bioreactor elements, this will act to limit the extent of aging even further, thereby helping to achieve lower effluent P. For example, specifying a velocity gradient of 300/s on the operation tab of the channel element, and a mixing power of 50 W/m3 on the power tab of each bioreactor (one anoxic and two aerobic) helps to reduce the effluent PO4-P concentration to 0.39 mg P/L. Note that these changes were made solely for illustration purposes; these are not practical values. Typical velocity gradients should be used to mimic expected mixing in activated sludge systems.

Impact of Dose Location
The results in the table below explore the impact of dose location on chemical P removal. Three identical configurations were evaluated with different metal addition points. The metal addition point varied between the influent channel, the aerobic zone, and a channel upstream of the final clarifier. The three flowsheets are shown below. Each configuration dosed 180 kg/d of ferric chloride. The influent channel is set up to provide a linear velocity of approximately 0.56 m/s and a residence time of approximately 1.2 minutes. Within the Operating tab of the influent channel’s property dialog box, the mixing at the point of metal addition is specified with a velocity gradient of 150/s. The mixing in the remaining zones is the velocity gradient associated with the power dissipated by the liquid flow (~115/s). The effluent channel is set up to provide a linear velocity of 0.24 m/s and a residence time of approximately 1 minute. The velocity gradient provided by liquid flow throughout the effluent channel is 76.36/s.

Metal-P interactions are dependent on the soluble P concentration; i.e. interaction rates increase with soluble P concentration since the driving force for removal is higher. As a result, upstream interactions are generally more dominant than downstream interactions since the P concentration decreases due to synthesis uptake moving downstream through a process. Based on the configurations explored, the effluent soluble P concentration increased as the dose of metal was moved further and further downstream and less bound HFO species are found with higher residual P concentration. The differences in unbound HFO species cannot be as easily compared since the channel elements provided different levels of mixing and hence different levels of HFO aging occurred.

Conclusions

This edition of the BioWin Advantage discussed setting up and simulating chemical P removal in BioWin version 6.x. We trust that you found this technical topic both interesting and informative. Please feel free to contact us at support@envirosim.com (Subject: The BioWin Advantage) with your comments on this article or suggestions for future articles.

The EnviroSim Team

Appendix A

Chemical Phosphorus Removal with Iron Salts

The following figure illustrates the processes involved in chemical phosphorus removal with iron salts.

When the model option Include ferric – phosphate adsorption/precipitation reactions is selected addition of ferric to water results in the rapid precipitation of hydrous ferric oxides (HFO) with a high number of active surface sites for interaction, HFO - High surface. When the model options Include ferric – phosphate adsorption/precipitation reactions AND Include iron reduction/oxidation reactions are selected, addition of ferrous to an oxidizing environment can result in the oxidation of ferrous to HFO – High surface.

Unbound HFO - High surface will co-precipitate with H2PO4- or H+, and depending on mixing, age to unbound HFO - Low surface (which has a low number of active surface sites for interaction compared to the high surface). Depending on the pH, H2P04 - can co-precipitate with HMO - High surface to form HFO - High surface with H2P04-. When the pH is low, H+ adsorbs onto HFO - High surface to form HFO - High surface with H+. The level of interaction between HFO – High surface and H2P04-, H+ depends on mixing, the pH, and the equilibrium concentration of P. The number of high surface sites available for interaction is a function of mixing and the Ferric active site factor(high).

HFO – High surface with H2P04- and HFO – High surface with H+ will also age to the respective low surface forms of these compounds i.e. HFO – Low surface with H2P04- and HFO-Low surface with H+ depending on their respective aging factors. Note: When HFO – High surface with H2P04- ages to HFO – High surface with H2P04- the surplus P remains bound and is tracked as P bound to Aged HMO.

Unbound HFO - Low surface will interact with H2P04- or H+, and depending on mixing, age to HFO - Aged (which has no available surface sites for interactions). Depending on the pH, H2PO4- will adsorb onto HFO - Low surface to form HFO - Low surface with H2P04-. When the pH is low, H+ adsorbs onto HFO - Low surface to form HFO - Low surface with H+. The level of interaction between HFO – Low surface and H2P04-, H+, or colloidal material depends on mixing, the pH, and the equilibrium concentration of P. The number of low surface sites available for interaction is a function of mixing and the Ferric active site factor(low).

HFO – Low surface with H2P04- and HFO – Low surface with H+ will also age to HFO – Aged. Note: The P associated with the low surface P-bound species will be tracked as P bound to Aged HMO.

If the model option Include metal salt – colloidal material coagulation reactions is turned on then colloidal COD will compete with H2P04- for active surface sites on the metal oxide thus lowering the efficiency of P removal. When the Metal-colloidal coagulation option is on some of the HFO – High surface and HFO – Low surface will be converted to HFO – Aged, making less available for P removal. In addition, CODp - slowly biodegradable colloidal is converted to CODp - slowly biodegradable particulate. CODp – slowly biodegradable particulate can then settle out of solution in a primary clarifier. The level of interaction between colloidal COD and metal depends on the mixing at the point of addition and within the activated sludge process, the residence time provided, and the level of competition between colloidal COD and soluble P for precipitation/adsorption.

Further information on the BioWin model parameters involved in chemical phosphorus removal with iron salts is found in the BioWin Manual.

Chemical Phosphorus Removal with Aluminum Salts

The following figure illustrates the processes involved in chemical phosphorus removal with aluminum salts.

When the model option Include aluminum – phosphate adsorption/precipitation reactions is selected addition of aluminum to water results in the rapid precipitation of hydrous aluminum oxides (HAO) with a high number of active surface sites for interaction, HAO - High surface.

Unbound HAO - High surface will co-precipitate with H2PO4-, and depending on mixing, age to unbound HAO - Low surface (which has a low number of active surface sites for interaction compared to the high surface). Depending on the pH, H2PO4- can co-precipitate with HAO - High surface to form HAO - High surface with H2P04-. The level of interaction between HAO – High surface and H2P04- depends on mixing, the pH, and the equilibrium concentration of P. The number of high surface sites available for interaction is a function of the mixing and the Al active site factor(high).

HAO – High surface with H2P04- will also age to HAO – Low surface with H2P04- depending on the respective aging factor. Note: When HAO – High surface with H2P04- ages to HAO – High surface with H2P04- the surplus P remains bound and is tracked as P bound to Aged HMO.

Unbound HAO - Low surface can either age to HAO - Aged (which has no available surface sites for interaction), or interact with H2P04-. Depending on the pH, H2PO4- will adsorb onto HAO - Low surface to form HAO - Low surface with H2P04-. The level of interaction between HAO – Low surface and H2P04- depends on mixing, the pH, and the equilibrium concentration of P. The number of low surface sites available for interaction is a function of the mixing and the Al active site factor(low).

HAO – Low surface with H2P04- will also age to HFO – Aged. Note: The P associated with the low surface P-bound species will be tracked as P bound to Aged HMO.

If the model option Include metal salt – colloidal material coagulation reactions is turned on then colloidal COD will compete with H2P04- for active surface sites on the metal oxide thus lowering the efficiency of P removal. When the Metal-colloidal coagulation option is on, some of the HAO – High surface and HAO – Low surface will be converted to HAO – Aged, making less available for P removal. In addition, CODp - slowly biodegradable colloidal is converted to CODp - slowly biodegradable particulate. CODp – slowly biodegradable particulate can then settle out of solution in a primary clarifier. The level of interaction between colloidal COD and metal depends on the mixing at the point of addition and within the activated sludge process, the residence time provided, and the level of competition between colloidal COD and soluble P for precipitation/adsorption.

Further information on the BioWin model parameters involved in chemical phosphorus removal with aluminum salts is found in the BioWin Manual.

Appendix B

The following figures illustrate the plug flow channel element, the Dimensions tab of the element’s property dialog box, and the Operation tab of the element’s property dialog box.

The channel element is essentially 4 bioreactor elements in series. When a metal addition element is connected to a channel element, the metal gets added into the first zone of the channel. On the Operation tab of the channel’s property dialog box, the velocity gradient at the point of metal addition (i.e. within the first zone) can be specified. The velocity gradient that gets applied to the first zone will be the maximum of the value specified or the velocity gradient associated with power dissipated by the flow of the liquid through the channel. The mixing in the remaining 3 zones is the velocity gradient associated with liquid flow.

The liquid depth and width of the channel should be specified on the Dimensions tab of the channel’s property dialog box. The liquid depth, width, and the flow through the channel are used to calculate the linear velocity. BioWin does not account for the hydraulic gradient in a channel. Typically, a linear velocity above 0.3 m/s is required to avoid settlement. In addition, the velocity in an influent channel containing raw wastewater should be higher when compared to the velocity in a downstream channel containing mixed liquor or effluent.

The length of the channel together with the depth, width, and flow through the channel will set the residence time of the channel. The residence time along with the mixing through the channel will determine the level of aging that occurs within the channel and hence the degree of interaction between HMO and H2P04-, H+ (ferric only), or colloidal material depending on the pH and what model options are active. It is recommended to adjust the channel dimensions so that a residence time of at least 1 minute is provided.

Note: When condensing multiple channels into one combined channel it is important to ensure that the width specified in the one combined channel is equal to the total width of all of the channels together. This will ensure that the linear velocity is representative of what is actually observed at the plant.