Preparing the Structure for Slag and Monitoring

The ISCO automatic samplers have finally arrived along with the flow monitoring equipment.   The flow and sample monitoring will be observed remotely at our office in Stillwater through the purchasing of a Verizon data plan.  Basically, we will be able to monitor what the samplers are doing/measuring at any time.  This way, when there is a flow event, we will know it immediately and obviously know to go to the site to collect the sample bottles as soon as possible for laboratory analysis. 

We recently poured some concrete at the inflow side of the structure.  The picture below shows the “before” picture and you can see that the inflow pipes (black) are at varying elevation from the surface, which is bare soil.  We wanted to create a clean and level “apron” where the inflow runoff water can enter the structure.  Eventually, there will be perforated plastic pipe attached to the other side (inside the structure) of the black metal pipe.  This pipe will be buried in the slag and will serve to evenly distribute the inflowing water (i.e. serve as a manifold).  

The next pictures show the concrete work:

 The finished product:

Also on this site visit we can to lay the pipe which will hold the suction lines for the automatic samplers and the bubbler tube for the flow monitor.

Not very exciting.  However next week we will sieve the slag and put it in the structure and finish up the installation of the monitoring equipment.

I would like to briefly highlight the work by Dr. Stefan Jansen in The Netherlands, here.

Flume installation and preparation for monitoring: Part 1

This portion of the construction and installation process for the P removal structure is NOT necessary for the typical-everyday use.  What is described here is for very detailed monitoring of performance for the purpose of research.  It is completely unnecessary for the applied purpose of the P removal structure.  We are doing this so that we can demonstrate the effectiveness of the unit and also produce more data for model verification.  This section is mostly relevant to researchers.

That being said, we will not put the PSM (treated steel slag in this case) into the structure until all the monitoring equipment is installed.  For monitoring, our purpose it to measure the amount (volume) of runoff treated by the structure during each runoff event and also measure the P concentrations flowing in (i.e. before treatment) and after (i.e. after treatment).  Knowing this, we can then estimate how much (mass) P has been trapped by the P removal structure. 

On the “inflow” side of the structure where water will enter, we will use an ISCO automatic sampler that will be triggered to collect samples based on the detection of flow at the “outflow” side of the structure where the treated water will be located.  Below you can see some photos of the construction of the approach for the 3 foot flume that will be used to monitor flow rate.  

Monitoring flow rate is critical if one is serious about getting a real estimate on the P removal performance of a P removal structure.  Many researchers have utilized only water sampling in their monitoring regime without the use of flow monitoring.  Simply put, P concentration testing of the treated water compared to untreated water alone is not sufficient for assessing the capacity of a P removal structure; knowledge of the flow rates and therefore the volumes of water treated is absolutely critical.  Any assessment without the use of detailed and thorough flow monitoring should not be taken too seriously.  Why? The short answer is “P load”.  If I measure a 100% P concentration reduction and there is only 1000 gallons of water treated, it is not correct to compare that concentration reduction to another scenario where there is only 50% P concentration reduction and 1 million gallons of water treated.  Simply put, the most important factor is the P load (i.e. mass) reduction.  Instead, it is more correct to compare the P load reduction (i.e. mass of P removed).

Too often, we focus on the final concentration of runoff water and treated water.  This makes little sense, especially when surface water quality thresholds (i.e. critical P concentrations for streams and lakes) are applied to the context of runoff water concentrations.  For example, runoff concentrations from a certain field (call it field 1) may test at 1 mg P/L, while field 2 may produce 0.2 mg P/L.  That does not mean that field 2 is somehow more “safe” than field 1.  Consider a scenario where field 1 produces 2 million L runoff for an event, while field 1 only produces 1000 L.  The P load transported off site for field 1 and 2 would be 1 g vs 400 g, respectively.  Concentrations alone would be deceiving in that case.  What matters is the mass or load of P that reaches the lake or stream; concentrations (mass/volume) change with dilution, evaporation, etc., but it is the mass of P that does not change.  This is why the USEPA has moved to a “total maximum daily loading”, or TMDL system for point and non-point source pollution concerning nutrients. 

For the same reasons, it is important to assess the performance of a P removal structure based on P load reduction, not P concentration reduction. 

Below you can see the flume installation process.  

This flume can handle much more water than our projected 2 yr storm (~16 cfs).  This flume is being used courtesy of Dr. Sherry Hunt, Kem Kadavy, and Ron Tejral, located at the USDA-ARS Hydraulics Engineering Research Unit.

All of the treated water plus any water that might overflow the structure will flow through the flume for measurement.  There will not be any water overflowing the structure if none of the storms exceed a 2 yr return period for that location.  In addition, if that does happen, there will be a flow sensor (actually a depth sensor) placed on top of the structure that will also monitor exactly how much water overflows the structure and remains untreated.  Below is the view looking from the structure downhill toward the flume:

This area between the outflow drainage of the structure and the flume will have an impermeable liner placed on the ground and “bordered” with stabilized railroad ties to force all water from the structure to flow through the flume where flow rate will be monitored and also where samples will be taken by the ISCO sampler. 

The photos below show the installation of the small building that will house the ISCO sampling equipment:

Construction and Installation of the Structure

Several posts back we determined how much PSM (treated slag in this case) was required to meet our P removal goals at this site, and we also determined how to orient that slag (i.e. area and depth) at our site to be able to treat all the runoff from a 2 yr-24hr storm event. 

Now it is time to build the structure.   In this case we are going with the low-tech, standard box structure where water flows through the PSM from the top-downward into subsurface drainage pipes.  With one small twist however: the drainage pipes at the bottom of the structure will not protrude through. 

We also designed our structure to be easily cleaned out with a front-end loader or a skid-steer. 

Below is the 33 x 13 ft structure from the perspective of downstream (drainage) side looking up to the upstream (entrance).  This was built in a modular form so that we can take it out into the field in pieces and assemble on-site.  Note the expanded metal on the drainage side.  The buried perforated pipes will drain to the expanded metal, where the treated water can then exit the structure.  This downstream side of the structure was designed as a “gate” to be removed when the PSMs become saturated with P.  At that point, the gate can be unbolted and easily removed, providing access for a skid-steer to drive in and scoop up material.

 Below is a picture of the “upstream” side where the runoff water will enter into the structure.  There will be perforated pipes connected to the metal pipes that will serve as an “entrance manifold” in order to evenly distribute the runoff water over the top of the PSM.

Here is most of the structure in pieces as we take it to the shop for painting:

Getting the primer on:

Then the paint:

 These students are talented!

Heavy, but not so heavy that we could not lift them by hand. Shown in the picture is Stuart Wilson – technician, Alexandre Ricardo Alves (i.e. The Shark) – Brazilian student intern, and Josh Daniel – graduate student.

Putting the pieces into our previously made “footprint”.  Note the adviser is actually working the shovel.

It was wet that day:

Completely put together: Note the earthen berms meet at the entrance to the structure:

Total cost for the metal materials and for a private fabrication shop to custom construct to our specification: $2,500.  Powerhouse, in Stillwater, OK.  405-377-6396. They did an excellent job.

Site preparation for construction of the P removal structure

As discussed in the post on site assessment, water must be channeled into a single location for a P removal structure inflow point.  That means that if you are working with a drainage ditch or a single subsurface drainage pipe, then it is pretty easy.  The flow in that case is already concentrated for you. 

For our site however, the flow was only somewhat concentrated along the gravel road in front of the poultry houses and on the East-West gravel road.  The flow from the field also combined with the gravel road into a natural drainage ravine.  We simply had to build some earthen berms to concentrate the flow into the P removal structure.  The figures below show the approximate location of the berms (in orange) and the P removal structure (in blue) for both an aerial view and a contour map.  The eastern berm was very short since it butted up against a steep hill, which acted as berm for channeling water.  

After the berm was constructed, we seeded with tall fescue and covered with an erosion-control mat:

Since then, the grass has been well established.

Also at this time, we cut the foundation for where the structure would be located.  As discussed on the previous post, we chose to use ¼” sized slag that was treated, which means that our structure required 40 tons of material, and would be approximately 33 ft long by 13 ft wide with a 20 inch depth of material.  This depth of material was chosen based on the slope of the site (see the contour map).  We simply cut into the ground 20 inches deep on the upslope side, and made a flat surface for 33 ft.  At 33 ft, the total change in elevation will be about 20 inches.  The cutaway view below shows the original ground surface relative to the “box shape” cut into the ground:

A word on drainage ditches

As discussed in the previous two posts, flow rate through the material is critical as it controls how much water will be able to be treated during a large runoff event.  Flow rate is controlled by the depth of the PSM, hydraulic conductivity of the PSM, and hydraulic head.  Hydraulic head is a function of the slope of the site.  Some sites have extremely low slope, and therefore, hydraulic head will also be naturally low.  This is the case on the Delmarva Peninsula where drainage ditches carry much of the surface water away from agricultural fields. 

Dr. McGrath has got around this problem by incorporating flow control structures to increase hydraulic head for the P removal structure, thereby increasing flow rate through the PSM and maintaining a more buffered and constant flow rate.  Our friends at Agri-Drain supply ideal flow control structures for achieving this task.  Essentially, the flow control structure acts as a miniature adjustable dam in the ditch.  The picture below show some of the P removal structures constructed by Dr. McGrath in drainage ditches and also some of the Agri-Drain flow control structures.

However, one must be cautious in the use of flow control structures to increase hydraulic head for the P removal structure because it can decrease the drainage upstream by backing up excessive water.  

Sizing the P removal structure part two. I know how much material I need: how should it be oriented to capture and treat the runoff?

The last post dealt with determining the required mass of your PSM for your site in order to remove a targeted amount of P in a given amount of time.  That mass of PSM could be oriented many different ways.  Two extremes are “wide and shallow” or “narrow and deep”!  

This is the part of the design that is really flexible and somewhat unique to the site.  Every site is different.  Different people will have different ideas for how to place the PSMs in the landscape filter.  However, the bottom line is that the water must flow through the material in an amount of time (i.e. retention time) that is sufficiently short enough to treat most of the water.  Any method that you want to do this is sufficient, as long as there is enough retention time for appreciable P sorption.  As discussed on the previous post, only Ca-dominated materials that are not buffered to a high pH seem to require a long retention time.  Gry Lyngsie at University of Copenhagen has some excellent data on that topic, using isothermal titration calorimetry. 

In other words, there is more than one way to skin a cat and I know that a lot of people are going to have some really great ideas for the actual physical layout on their particular sites.  Many of our European friends have demonstrated some really good ideas for moving the water through the PSMs.  Some folks like to design the water flow from the bottom of the sorption bed upward, laterally, or from the top downward.  There might be reasons at a particular site to choose one flow direction over another.  One reason to design flow movement from the bottom to the top is that you have the advantage of gravity in limiting the loss of the solid PSM material as they tend to stay settled in that situation.  A disadvantage to this type of design is that it often results in the PSM being saturated with water all the time, which could be a problem with Fe-dominated PSMs that can dissolve under reducing conditions.  An advantage to flow design from the top downward is that it is free-draining and avoids saturation with water during non-flow events. 

Regardless of the water flow direction through the material, flow rate is not faster for the top-down, bottom-up, or lateral direction.  For each of those situations, the flow rate is dependent on the hydraulic head, thickness of the PSM layer, and the hydraulic conductivity of the PSM.  In any of those situations, the standard Darcy Equation can be used to design the structure after you have determined the required mass of PSM, peak design flow rate, and site limitations such as area and slope (i.e. hydraulic head). 

Most of the structures that I build are designed to move water through the PSM from the top-down into buried subsurface drainage pipes, as shown in the diagram above.  This allows the structure to drain freely after the flow event has ceased.  Other folks do it differently.  Dr. McGrath also builds most of his structures to drain from the top down with subsurface drainage pipes.  One exception is his metal “boxes” that consist of a box made of expanded metal with a perforated drain in the center; the PSM is placed in box between the walls and the perforated pipe that drains it.  Such a condition would require some type of a radial flow equation to predict flow rate.    

How to design an appropriate layout that is suitable for the site limitations and peak flow rate

 Usually, the most limiting factor in structure design is the hydraulic conductivity of the PSM.  The dichotomy is that PSMs which have the best P sorption ability tend to have poor hydraulic conductivity, and PSMs with large hydraulic conductivity have low P sorption ability: 

Using a material with a low hydraulic conductivity translates to designing a structure that has a much larger area, since the thickness of the sorption bed must be lower in order to achieve a reasonable flow rate.  This is not a problem if your site is not limited in available area. 

Determining the layout of the structure for a particular PSM is a function of the following parameters:

1. Required mass of PSM
2. Hydraulic conductivity of the PSM
3. Porosity of PSM
4. Bulk density of PSM
5. Target peak flow rate for the structure
6. Maximum area for structure at site
7. Maximum hydraulic head at site

    Consider the site in which this blog is dedicated to (OK poultry farm): we have several PSMs available to us as described in the previous post in which a given mass will remove a certain amount of P.  However, some of the materials have a low hydraulic conductivity.  The table below shows the potential layout for several PSMs that we had to choose from.  Note that this table was developed specifically for our site characteristics, and the ability to handle the flow rate expected for a 2 yr-24hr storm event (16 cubic ft/s).

From the table above, it becomes apparent that the PSMs with lower conductivity tend to have a greater P sorption ability, as those materials (WTR, AMDR, and fly ash) all require relatively small amounts of PSM.  However, those materials also tend to require large areas.  On the other hand, Use of the sieved steel slag also requires a large amount of area, not because of limited hydraulic conductivity, but because of the physical constraint of housing a large mass of material.   For our site, we chose to use the treated slag since it was a “happy medium” between the low hydraulic conductivity-high P sorption materials such as WTR, AMDR, and fly ash, and the high hydraulic conductivity-low P sorption materials such as the sieved steel slag.

The suitable layout for the different PSMs shown in the above table was estimated using the software that we developed for designing P removal structures.  Please contact us if you are interested in obtaining a license.     

Sizing the P removal structure: what is the required mass of P sorbing material (PSM) necessary for the job?

With the following information, we can now take the next step and estimate how much PSM is necessary for the job:

1. Annual P load at the site
2. Typical dissolved P concentration in runoff (or drainage) water to be treated
3. P removal goal i.e. the % of the annual P load that is desired to be removed
4. Characteristics of the locally available PSM

Annual P load and typical dissolved P concentrations

The annual P load was already determined as described in the previous post.  Turned out that at our site, we expect an annual P load of 22 kg.  This was calculated partly by knowing that our typical high-end dissolved P concentration is 2 mg/L.

P removal goal

Based on the large P load at this site, I am setting a range for removing the annual P load from 20 to 50%.
 

Characteristics of the locally available PSM

This is the part that requires a bit of work.  In order to properly design a P removal structure, one must develop a “design curve” for the PSM that they will be using in their structure.  A design curve is simply a quantitative description of the relationship between dissolved P loading of the PSM and the % discrete P removal.  This must be determined in a flow-through setting; a batch P sorption experiment will not suffice!  See Penn and McGrath (2011) and Stoner et al. (2012) for examples of this comparison.  We and other researchers have determined that in the context of a P removal structure, any assessment of P removal must be in a flow-through setting.  A batch sorption experiment in this context is only useful as an index to compare different PSMs, not to quantify how much P it will remove from a flowing solution.  There are several reasons why the batch sorption experiments fails in application to a flowing context such as a P removal structure:

1. Batch tests typically has a non-representative long contact time
2. Batch tests do not have constant removal of reaction products
3. Batch tests do not have constant addition of reactants (i.e. P) 

A design curve is specific with regard to the retention time (i.e. contact time) and the inflow P concentration that is moving through the PSM.  Below is an example of a design curve for an acid mine drainage residual (AMDR) that we collected from Eastern OK:

Note that this design curve is specific to an inflow P concentration of 1 mg P/L and a retention time of 30 seconds.  In other words, it takes 30 seconds for the inflowing solution to pass through the AMDR.  As you would expect, the P sorption is initially very high, but with further P loading the AMDR is not able to sorb as much P as it did previously.  The shape of this curve varies between PSMs, retention times, and inflow P concentrations.  This is discussed in detail in Stoner et al. (2012).

Knowledge of the equation of a design curve is necessary for choosing the necessary amount (mass) of PSM for your structure.  This curve can serve several purposes, depending on what you desire:

1. Estimate the lifetime of the structure if a given mass of a specific PSM is to be placed in the structure.  In this case, “lifetime” is defined as the amount of time until the P removal structure is no longer able to sorb P that flows into it.
2. Upon integration of the design curve, estimate how much P will be removed by the structure during that lifetime. 
3. Upon integration of the design curve, estimate how much of the PSM (i.e. mass) will be necessary to remove a desired amount of P under the condition of the design curve.     

For our purposes at this particular site, our use of the design curve will be for purpose number 3.  For example, at our site we have several PSMs locally available to us: steel slag from Fort Smith, drinking water treatment residuals (WTRs) from Tulsa, fly-ash from Muskogee, and an AMDR.  All of these are potential options for us.  Other PSMs may be available in different regions.  Some areas may have a greater abundance of steel slag or WTRs.  Perhaps you live in Western PA (God’s country) where AMDRs are highly available.  There may be some PSM that is unique to your region.

Considering our conditions previously determined: annual P load of 22 kg, typical inflow P concentration of 2 mg/L, and appropriate design curves for each material, it would require:

• 5.5 tons of the Tulsa WTR to remove 37% of the annual P load
• 3.1 tons of our AMDR to remove 49% of the annual P load
• 131 tons of sieved steel slag (1/4” and greater diameter) to remove 21% of the annual P load
• 41 tons of a treated sieved steel slag (1/4” and greater diameter) to remove 45% of the annual P load

Model for predicting the design curve instead of directly measuring it

We developed a model for predicting the equation of a design curve for a specific PSM, as a function of the retention time, inflow P concentration, and selected PSM characteristics (chemical and physical).  This model was developed for the following reasons:

1. Conducting a flow-through experiment for every single individual PSM sample and every possible flow condition (i.e. inflow P concentration and retention time) is extremely time consuming and expensive.
2. There is great variation in P sorption behavior between different PSMs.
3. There is great variation in P sorption behavior the same type of PSMs that come from different sources.
4. There can be variation in P sorption behavior among the same type of PSMs that come from the same source, but are produced at different times.   
5. It is much easier and less expensive to measure certain chemical and physical characteristics of PSMs to use in predicting a design curve, than it is to conduct many flow-through P sorption experiments. 

Through the tireless work of a former graduate student, Dustin Stoner (a living legend as far as I am concerned), we have developed a model to predict the equation for a design curve for a specific material under specific flow conditions.  One thing that we learned from conducting > 1000 flow through P sorption experiments, is that under the relatively short retention times that we are working with in the world of runoff and subsurface drainage P removal structures (seconds up to 20 minutes), retention time usually has little impact on P removal (see Stoner et al., 2012).  This is true for materials that dominantly remove P via fast kinetics by Al and Fe sorption (ligand exchange) and for Ca-rich materials that have a relatively high pH AND are buffered to a high pH.  For example, flue gas gypsum is an example of a Ca-rich material that is NOT highly buffered with regard to pH, and therefore the retention time does have a dramatic impact on P removal in a flow-through setting.  Gypsum is one of the few materials that I have observed this for.  For most PSMs, retention time does not have much impact at our scale of retention time that we are aiming for.  However, for materials such as gypsum, increased retention time will greatly increase P removal (see McGrath et al., 2012).  

While the details of this model for predicting the design curve will not be discussed here, the model is at the heart of the current software that we are developing, which essentially helps one to design a site-specific P removal structure in the same manner in which this blog describes.  If you are interested in a license to that software, please contact me.  

Site data collection required for structure design

In addition to the estimates of dissolved P concentrations in runoff at the site location, further information was required:

1.   Peak runoff flow rate
2. Average annual flow volumes and dissolved P load
3. Hydraulic head

The average annual flow volume and peak flow rate was calculated by using site information required for estimating the NRCS curve number (ftp://ftp.wcc.nrcs.usda.gov/wntsc/H&H/training/runoff-curve-numbers1.pdf).  This required watershed information regarding soil type (used to determine hydrologic soil group), ground cover, greatest length of flow, and slope.  Each of these parameters, except for soil type and flow length (determined by soil survey and contour maps) were determined via site visit.  The wonderful NRCS folks at the local Stilwell office (Chad Kacir, Andy Inman, and others) were responsible for conducting the site survey for the curve number method and also calculations for peak flow rate and average annual flow volumes.  

Peak runoff flow rate estimate

The curve number is used in conjunction with the precipitation depth for the design storm.  In our case, we wanted to design our structure for a 2yr-24hr storm, which produces about 4 inches of rainfall as estimated by standard USDA rainfall tables (http://www.nws.noaa.gov/oh/hdsc/PF_documents/TechnicalPaper_No40.pdf).

For our site, the curve number (CN) was 78; the cover was mostly pasture.The curve number method resulted in a runoff depth of 1.97 inches for our 9 acre watershed, predicted for a 2yr-24hr storm.  Next, the runoff depth is used to calculate the peak flow rate through use of the Soil-Cover-Complex method and “time of concentration”.  The time of concentration is calculated using the curve number and the previously determined greatest length of flow via curve number method.  For our site, the greatest flow length to our proposed structure location was 1050 ft, which resulted in an estimated time of concentration of 24 minutes.  Based on this time of concentration, the estimated peak discharge was 0.9 cfs/acre-inch runoff.

For our scenario:

0.9 cfs/acre-inch runoff  * 9 acres * 1.97 inches runoff = about 16 cfs peak runoff rate.

Therefore, our goal was to design a structure that could handle a flow rate of at least 16 cfs so that it would be able to treat all of the runoff produced from a 2 yr-24hr storm.

Average annual runoff volume and P load estimate

Next, the annual flow volume is necessary in order to determine what the annual load of dissolved P is.  To achieve this, the runoff coefficient method was used.  An example is shown here: http://watershedmg.org/sites/default/files/docs/wmg_calculating_runoff_worksheet.pdf ) This calculation is simply based on cover for estimating the runoff coefficient, watershed area, and average annual rainfall depth.  For our site, the annual rainfall depth was 44 inches.  Based on our site, the average annual runoff volume is 12 inches/yr, or 9 acre-ft. 

Our grab samples indicated dissolved P concentrations between 1 and 2 mg/L.  Using the higher value of 2 mg P/L for over-design, this resulted in an estimated load of 22 kg of P/yr transported in runoff from the watershed. 

Therefore, our estimates for the required mass of PSM for this site will be partly based on the predicted value of 22 kg dissolved P lost in runoff per year. 

Hydraulic head

Hydraulic head is critical for achieving the desired flow rate through the P removal structure.  As discussed in the previous post, the runoff water must flow through the structure in order for P removal to occur.  Simply put, the hydraulic head is the force that “pushes” the water through the structure. The hydraulic head can be estimated by conducting an elevation survey of the proposed structure location. 

Based on the elevation survey, we will have an appreciable amount of hydraulic head to achieve the desired flow rates (i.e. 16 cfs for a 2yr-24hr storm).