Hello everybody! Welcome, and good afternoon. I’m Cheryl Reynolds, with the UC Statewide IPM Program. Welcome to today’s UC AG Expert Talk on understanding effective citrus spray application through computer simulations. Petr Kosina is here with me, and he’s going to be running our polls and doing some troubleshooting. So now I’d like to introduce our speaker today. Dr. Peter Ako Larbi is an Assistant Cooperative Extension [Specialist] of the Kearney Agricultural Research and Extension Center. He’s specializing in agricultural application engineering, and today he’s speaking on understanding effective citrus spray application through computer simulation. And so now I’m gonna pass this over to Peter. So hello, Peter, and you can go ahead and share your slides. [Peter:] Alright, good afternoon, everyone. Thank you for your participation in this webinar. I’m excited to present this webinar on understanding effective citrus spray application through computer simulation. Let me begin by giving you just a quick overview of my program. Being in my position as Assistant Cooperative Extension Specialist for about a year now, my responsibility is to provide regional leadership in the area of agricultural engineering extension, and applied research with the focus on spray application engineering. That’s my primary goal. I seek to improve agricultural productivity while reducing the impact of pesticides and other agricultural chemicals on the environment. There are five themes that guide my program, which are 1) evaluating spray application techniques, 2) testing commercial spray application technologies, 3) promoting best practices for safe, economical, and environmentally sound pesticide spray application, 4) developing novel spray application technologies, and 5) deploying spray application decision support systems. My program website is still under construction, and will provide more information on these themes. At the San Joaquin Valley area of California, where my program is headquartered, is a major production region for almond, pistachio, stone fruits, citrus, and grapes. And also where air blast sprayers are the main types of sprayers used to apply pesticides. For effective and economical pest control, it is critical to achieve high on-target deposition and coverage, but this can be very challenging and costly due to variability in three characteristics, sometimes within the same orchard. One way by which we can improve air blast spray application is really by understanding how it works. Here is the conventional air blast sprayer with a typical polar jet design. By its name, an air blast sprayer simply blasts out a mixture of air and liquid in the form of tiny droplets. The high- volume, high-velocity air serves as a vehicle to transport droplets. We may refer to this mixture of air and liquid as a spray cloud. So here is an air blast sprayer that is blasting out some spray cloud, and there is the nurse tank blocking the view, so you cannot see the outlet, but you can use this as the…the idea. There is an outlet and also there are nozzles lined up that spray the tank liquid. So we can see how far and high the spray cloud is traveling. And if I were to place a tree in a path of this spray cloud, part of this spray would be cut by the tree canopy. And this tree canopy would be my target tree. And I would be interested in knowing how much of this spray the canopy captures. Here are some basics of the application process. Both tank mix liquids are atomized, expanding to tiny droplets by a process of atomization, while exiting nozzles, which are under pressure. Now, the air transports the spray droplets on towards a tree canopy, which is the target tree canopy, and the tree captures part of the spray, leading to canopy deposition, which is our desired outcome from the application. Part of the spray that misses the canopy drifts beyond the canopy, and part of it would also deposit on the ground, either directly or indirectly through runoff from leaves. Now by maximizing canopy deposition, we minimize drift and ground deposition. Atomization produces very consistent droplets that have different sizes. The factors that affect the droplet size [are the] physical properties of a tank solution, such as density and viscosity, also we have nozzle design and operating pressure. As a spray cloud moves, there is an influence of weather factors. For instance, a high temperature combined with a low relative humidity, distance, [and] droplet evaporation favor spray drift. We already know the effect of high winds that also favor spray drift. At this point, I’ll post our polls— three questions. So the first question is which of the following is true about air blast sprayers? The second one is, “At any instance during an air blast spray application, which of the following defines the targets trees? And then the last, the third one is, “Which of the following is a desired outcome of an air blast spray application?” [Cheryl:] Okay, we’re going to be closing out this poll, and then you should be seeing the results screen. [Peter:] Okay, so I think most of you got one, two, and three correct. The correct answer is C for Question 1, and then it’s B for Question 2, and then that is A for question 3. Okay. The amount of spray that deposits on the tree canopy, [or] what we call the “on-target deposition,” is often lower than we desire. and this is because of the complex interaction between these factors: equipment design, application parameters, spray physical properties, tree characteristics, and weather conditions. Research showed that for 100% of material sprayed in citrus applications, only about 73% to 80% deposits on the canopy; 6% to 14% potentially drifts away from the application area, and 9% to 20% falls to the ground. However, these estimates were based on limited research data and may not apply to the direct situations that may prevail during an application. So the question remains, how can we estimate these values, that is the canopy deposition, the potential drift, and potential ground fallout or ground deposition? How can we estimate these values for planning and evaluation purposes for the wide range of situations that could exist, such as different tree characteristics, different orchard conditions, [the] right combination of application parameters, and a host of weather parameters? You and I know that field experiments present major limitations in terms of time, labor, material, and other resources. Nonetheless, computer modeling and simulation can help us overcome these limitations at a low cost. So this was part of my motivation for my Ph.D. work some eight years ago, in which I developed a model to predict spray deposition in citrus applications. Now, by model, I do not mean any of these or their likes, but I mean a bunch of mathematical formulas describing the various processes of carrying a spray cloud, from when it is generated, to be on the target tree canopy. So here in the top-left is what we call a forest diagram of the spray system, which shows different processes happening concurrently within the spray cloud, as it disperses along. Next here is a conceptualization of the spray cloud, where it’s partitioned into compartments of equal thickness. And each compartment is characterized by cross-sectional area, volume, and air velocity. And in this plot, this plot shows an increasing cross-sectional area with a corresponding decrease in air velocity as the spray cloud moves away from the sprayer outlet. We see that the air velocity drops sharply as it exits the sprayer outlet, and then gradually decreases further away from the sprayer. For the simulation, the spray generated was grouped into ten categories, from extremely fine to extremely coarse, and I consider the droplet evaporation, or carrying, over the compartments, and then also, over time. So finally here is a schematic representation of spray compartments outside of a canopy, and then inside of the canopy. You can see the various processes of carrying within each compartment, from which the model equations were formulated. So, we have an equation for the compartments outside the canopy, an equation for the compartment inside the canopy. This is what was solved by simulation to generate a spray outcome. Here are some simulation results. First is sprayer air velocity, or different air flow rates. We see that the air velocity, generally…the air velocity generally increases with increasing air flow rates. Now next here, shows [the] droplet size spectrum for sprays from different nozzle sizes. These are the nozzle sizes by their colors. Sprays can be characterized by volume median diameter (VMD), which is the droplet size at which half of the spray volume is made up of droplets smaller than that size, and half is made up of droplets larger than that size. So the volume median diameter, VMDs, are depicted here in the drop-down lines, and they summarized in this table for the different nozzle sizes. At the bottom here, we see the effect of operating pressure on [the] droplet size from a single nozzle. Increasing pressure decreases droplet size. You can see how it’s depicted here. This shows a a clearer picture of how the droplet size decreases with increasing operating pressure. Here on the right is a simulation of droplet evaporation over time. So each of these lines represents a different droplet size. We see the change in droplet sizes over time with the smallest droplet size evaporating out of the spray faster, and the next size would follow suit as elapsed time continues to increase. Down here shows the changing percentage of active ingredients as the water evaporates out of the droplets. You can see how quickly the concentration, or percentage of active ingredient, changes based on the droplet size. Now let us look at the effect of weather on the airborne spray mass as it moves away from the sprayer outlet. We have here, air temperature, relative humidity, wind speed, and wind velocity— actually wind direction, sorry. Now in brief, at any given distance away from the sprayer, the airborne spray mass decreases with increasing temperature, and also decreases with decreasing relative humidity, and then increasing with [decreasing] wind speed, and then also increasing [with decreasing] wind direction. Where ninety degrees represents a right-angle cross-wind, or the spray cloud direction. Alright, let’s look at a couple of poll questions. The first one is “Which of the following reasons incorrectly justifies the need for or use of model simulations?” The second is, “Which of the following weather conditions should be avoided because of its effect on spray application?” [Cheryl:] Okay, I think we’re gonna bring up the results now. [Peter:] Okay. Alright, I think that shows that many of you, I mean 64% of you, got it right, you got the first one right. And 93% got the second one right. So many of you are scoring these poll questions. Well, for the first one, the correct answer is C, “model simulations can create cool graphs that cannot be created with actual field experiments.” That’s really not what we aim at by using model simulations. And for the second question, [the answer] is C, “high wind speed, because it favors spray drift.” So you want to avoid high wind speeds at all costs, to improve your spray application. Alright, let’s carry on. So, now in order to build confidence in the model, I validated the model with two field experiments— one for dispersion, which is just looking at the spray cloud dispersing from the sprayer outlet, away from the sprayer outlet. And then the second test looked at spray cloud interaction with the canopy to result in deposition. This was done in a commercial orchard. Here are some results. The markers are the field data, and the lines are the model simulations. Without going too much into details, the model efficiency was 78% for dispersion, and 61% for deposition, which were considered very good for such a complex model. So I pushed further to develop an expert system, with a graphical user interface to answer the outstanding questions: How much spray deposits on the canopy? How much spray potentially drifts? And how much spray falls to the ground, or deposits on the ground? And I call that expert system “Citrus Spray Ex.” So here is a structure of the expert system, with the arrows showing the direction of information flow. It consists of a knowledge base, and an advice module, both based on technical literature, which includes direct expert knowledge. The inference module receives input from a user through the graphical user interface, interacts with the knowledge base and advice module, running the spray model as necessary, and then retains some results to the user. Here is the simplified flow chart of the spray evaluation simulation, the path that runs the model. And here is a graphical user interface for the evaluation simulation part of the expert system. The user inputs mainly consist of application parameters, tree characteristics, orchard condition, and weather parameters. And the outputs are canopy deposition, ground fallout, and potential spray drift; all in percentages. There is also a “What If” pane, which tells the user what changed from a previous run, and how it affected the output. And below here, is the result of evaluating the expert system. Here is one quick poll question to answer. It states, “Validating a model with data from an actual field experiment gives us some confidence to try the models predictions or make decisions based on it— true or false?” [Cheryl:] I think we have most people responding, so the results should be coming up. [Peter:] Okay. Thank you very much for the response. I think we all agree, 100% responded correctly. So we’ll move on. So, now let’s look— now use the expert system to learn something about different variables, without getting into the simulations that uphold well on changes in the variable in question and how it affects the output. So let’s begin with air flow rate. Apart from transporting the spray droplets, the air helps in penetrating the tree canopy. Now this is what we…this is what we see when we change the air flow rate from 19,000 CFM to 35,000 CFM, and then to 48,000 CFM. Canopy deposition increased from 45% to 61%, and to 62%, with some increase in ground fallout, while spray drift reduced from 48% to 28%. Please take note that these percentages of the total volume applied, as shown in the table above, in subsequent slides, changes in the volume applied will be highlighted, so please take note of that as we move on. So how about nozzle size? We know that increasing nozzle size, that is, increasing nozzle size increases the volume of spray being discharged, as you can see here, and [increases] the droplet size. So both the spray volume and the droplet size increase by increasing the nozzle size. We learn here that increasing nozzle size from D3-13, and these are columns for D3-13, to D4-23 and D5-25, increase and then decrease deposition as a percentage. Look at the total volume applied here to better appreciate these percentages. So, increased and decreased canopy deposition as a percentage…and we can see that there is a trade-off here between drift and ground fallout. Drift decreases from 34% to 21%, while ground fallout increases. And this would most likely be due to spray runoff from leaves, as the spray volume increases. How about operating pressure? Increasing operating pressure also increases the volume of spray being discharged, and tends to decrease droplet size. Here we learn that increasing operating pressure from 115 PSI to 175 PSI decreases the percent deposition in favor of spray drift. Let’s take this poll question: “According to model simulation results from preceding slides, which of the following general statements about citrus air blast spray application is true?” Okay, we had a few misses here, I mean quite a few here. So, according to the model simulation results from preceding slides, which of the following general statements about citrus air blast spray applications is true? We have, “Increasing air flow rate increases percentage canopy deposition,” We found that to be true in preceding slides, so the correct answer is A. “Increasing nozzle size increases percent deposition and percentage potential spray drift,” is false from the previous slides. Spray drift actually decreased in favor of of ground fallout. Alright, let’s move on. Let’s look at a few more variables. Let’s look a ground speed. You probably know that generally increasing ground speed…increasing ground speed decreases application rates, that is your GPA (gallons per acre). Here we also learned that increasing ground speed increases percentage canopy deposition, at the expense of ground fallout and spray drift. And you need to bear in mind that the volume, the total gallons being applied, is reduced, but there is an increase in the percentage deposition and there is a decrease in potential spray drift, and also ground fallout. How about canopy density? That is, the canopy foliage density? How does this affect the application outcome? Here there is no change in volume applied, so we are just looking purely at the effect of canopy deposition— or, sorry, canopy foliage density. So having more leaves in the same canopy volume increases spray interception, leading to increased canopy deposition, at the expense of spray drift. Although there is potentially some increase in ground fallout, which would occur as a result of some runoff from leaves. And now let’s look at— talk about weather. How do higher relative humidity values affect the outcome? We learned here that increasing relative humidity increases percentage deposition, so increasing relative humidity from 60% to 90% increases percentage deposition from 56% to 66%, because there is low droplet evaporation as a result of increasing the relative humidity, spray drift suffers in this case. Finally let’s look at wind speed. This is looking at crosswinds. That is, wind or carrying 90 degrees to the direction of the spray cloud. So we see here, that increasing wind speed actually reduces the canopy deposition in favor of spray drift. So this just emphasizes what we know about the effect of wind speed, and the crosswinds just represents the worst case scenario. Let’s look at this poll question: “According to model simulation results from preceding slides, which of the following general statements about citrus air blast spray applications is true?” That is from a few preceding slides. [Cheryl:] And while they’re answering, there was a question that came in for you, if I can ask you that. Okay so, “In the model simulation, does drift eventually become ground deposition, but not measured in the simulation as such? So in other words, does ground deposition mean that the pesticide hits the ground in the adjacent row?” [Peter:] If I understand your question correctly, will the drift beyond the target canopy, will it eventually land on the ground, or land on adjacent rows? Is that correct?” [Cheryl:] Yeah, I think so. [Peter:] Okay, yes, this drift here is potential drift, so it’s evaluated just at the exit point of the target canopy, so it’s not looking at where it potentially falls, but in reality, part of that spray would end up on adjacent rows. While, you know, a majority, depending on the conditions in the field, part of it would also drift away from the the application site altogether. But it is really difficult, to some extent, to evaluate where the spray actually results beyond the target. So, [the] best practice would be to keep the spray within— I mean up to, the target canopy, if possible. Work within the target canopy, because there is uncertainty in whatever misses the target. [Cheryl:] Okay, thank you, and I think Petr has just brought up the results. [Peter]: Okay, yeah, I think a majority of you responded correctly, which is, “According to the model simulation results from preceding slides, which of the following general statements about citrus air blast spray applications is true?” A is the answer. “Increasing sprayer ground speed increases percentage canopy deposition.” That is a correct answer from the simulation results. Okay, let’s move on quickly. I think I’m falling little behind schedule. I’ll bring up some other applications that we have used as a model to address and I think this would be useful for our purposes here. So, we have used the simulation runs from this expert system to test some advanced air blast spray systems, to determine the benefits for not just spray material savings, but also deposition savings. We focused on four advanced concepts. One is automatic nozzle rate adjustment, or that’s nozzle flow adjustment, automatic air assistance control, automatic application rate control, and automatic nozzle on/off, which in certain nozzles, off when [inaudible] it’s needed, and the physical manifestations of these concepts take different forms. Here are some default settings we consider for the simulation, and for lack of time, I will not go, or not delve much into this, but I just want to show you how these compare. So here are the configurations of the different spray concepts for different tree sizes. We considered three different tree sizes. A small size, medium size, and a large size. And also, three different foliage densities: low, medium, and high. The conventional application was configured to the large tree, and applied at 68 gallons per acre for all tree sizes. The air flow rate was set to 38,000 CFM. The automatic nozzle rate adjustment system was configured to apply a 44% nozzle rate for small trees and a 62% nozzle rate for medium trees, yielding 20 gallons per acre and 34 gallons per acre, respectively, but using the same air flow rate as a conventional application. The automatic air systems control— this particular one varied the air assistance amount. For small-sized trees, you can see, 18,000 CFM for small-sized trees, 28,000 CFM for medium-sized trees. And then the automatic application rate control was tested within 10% of states from the desired speed of 2 miles per hour. So the same application rate was achieved due to correction, correction by this particular system. Slowest speeds were corrected by reducing pressure and thereby reducing volume. Faster speeds were also corrected by increasing pressure to increase the volume to match the desired application rate. And for the last system, automatic nozzle on/off control, the sprayer was configured to use different configurations of nozzles for the small- sized trees and the medium-sized trees, yielding 16 gallons per acre and 40 gallons per acre for the small and medium-sized trees. And then we, here is the configuration, the tree configurations used. We had three uniform configurations and we had three non-uniform configurations, combined with with three foliage densities. You can see a depiction of it, percentage small trees, percentage medium trees, and percentage large trees. So you can see how these compare visually. And then for data analysis, we focused on just deposition the canopy deposition, and ignored the other losses, which are ground fallout and spray drift. For comparison among the various treatments, we looked at volume applied to a number of trees to get a spray rate, and the volume deposited divided by number of trees [to get the] deposition rate. And spray savings is a deficit volume sprayed compared to corresponding conventional air blast applications where these advanced systems are not being used, and also deposition savings was evaluated as supplementary deposition gained compared to corresponding conventional air blast application. So total savings combined two savings in this case. So here are some results here. The same volume of spray was applied for the conventional spray application of the ground area, but we found out the spray rates increased with percent missing trees, and also decrease with increasing travel speed. So increasing percent missing trees and decreasing travel speed lead to negative savings in each of these cases that were tested for the conventional application. We found that the deposition rate was limited by the canopy size and also foliage density, and then compared to the uniform large tree with high foliage density, deposition savings also varied with the three configurations, but were all negative. Also, deposition rate and deposition savings reduced with increasing travel speed. This establishes the need for a variable range of techniques. For the automatic nozzle rate adjustment, only nozzle rate was adjusted, but not nozzle configuration. So ten nozzles a side were used for all the tree sizes. So spray rates varied with tree size, but reduced by the reduced nozzle rate led to reduced canopy deposition rates, and then also negative deposition savings. For the automatic air assistance control system, only air volume was adjusted, but not nozzle rate or nozzle configuration. So there was no spray savings, but the reduced air led to reduced deposition rates and negative deposition savings compared to conventional application. For the automatic application rate control system, application rate was maintained at a conventional rate, leading to positive spray savings at lower speeds, and negative savings at high speed, but this is actually an adjustment, so it may not necessarily be viewed as negative savings. Deposition rate was fairly adjusted in this treatment, although there was still some deficit at high speeds than at the target speed. For the automatic nozzle on-off control system, only nozzle configuration was much too precise, and no spray was discharged in tree gaps. So there was no deposition savings per se, but there was much savings in the spray rate. So overall, this is how it looks. Automatic nozzle on-off control system seems to give the highest benefit followed by automatic nozzle flow control system, and then the automatic application rate control system. The air assistance control system did not show any benefit, likely due to their great distance of a canopy from the sprayer’s outlet. So one thing we can glean from this is that [inaudible] sprayers can benefit from combining multiple concepts to optimize both spray savings and deposition savings for this economy. Let’s look at these quick poll questions as the last poll questions. And then we’ll round up. “From simulation results in earlier slides, which of the following advanced systems…which of the following advanced systems may benefit (in terms of material savings) spraying an orchard having uniform foliage density hedgerows— that is, the tree canopies touching— and also trimmed at the top and sides? And the second question says, “In an orchard with variable tree sizes, uniform canopy gaps, and possible missing trees, which of the following advanced systems may provide the greatest benefit in terms of material savings? [Cheryl:] Okay, just a few more seconds, and we’ll bring up the results panel. [Peter:] Okay. I know this particular one was a bit tricky, because you need to read carefully to pick the answers. For the first one, the correct answer is B; that’s “An air blast sprayer with an automatic application rate control” in this case, because you have uniform foliage density, and you also have a uniform tree size and all that. So none of these would benefit greatly— the others would not benefit greatly. Okay. And then for the second one, the correct answer is C, “an air blast sprayer with automatic nozzle on-off control based on the presence and absence of trees,” because you possibly have some missing trees, and you also have gaps between trees. Okay, thank you very much. So some take-home messages here…we know that guess-timating the outcome of an application, an air blast spray application, that is, in terms of canopy deposition, drift, and ground fallout…is almost impossible because of all the complex interactions. Using modeling and simulation tools for predictions can improve decision-making for better planning. And in this presentation, I have shown that the Citrus Spray Ex expert system, which was using all these simulations, or similar tools, can help decisions, provide decision support for these purposes. And I want to quickly— and I’m sorry I used up all the time, but I really wanted to show you this upcoming opportunity because I think, especially for growers and applicators this could be something that would benefit you. We have a project that is being funded by DPR. It’s just about to begin. It’s a two-year project, and this would be evaluating the expert system by end-users. We would have a project website providing information, providing installation and tutorials and a host of other information. We’ll be having workshops, hands-on calibration trainings, and field days. I want just to throw it out there, so if you seem to be interested, I would work through the organizers of this webinar, to reach out and see if you are interested. You can sign up for it and the information will be coming out later to industry. So people can sign up for this evaluation, and possibly attend some of our meetings, and gain insight into using this expert system. So thank you very much. Here are some of the references that I used in the presentation, and I welcome any questions if there are [any]. [Cheryl:] Thank you, Peter. So it doesn’t seem like we have any questions, but Peter, thank you again for presenting today. [Peter:] Thank you very much for having me. [Cheryl:] And we want to thank everybody for attending.