Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.basic.visual.misc > #3267

Process Control By K Krishnaswamy Pdf 27

Show all headers | View raw

His research expertise are on studying soil physical, chemical and biological properties in relation to nutrient cycling and soil quality in wetland and upland systems. Utilizing microbiology as a central tool, we conduct research on microbial structural and functional diversity by traditional and molecular approaches, environmental influence on microbes structure and function, pesticides fate and transport, apply enrichment techniques to isolate and characterize pesticides degrading microbes, biodegradation processes in soil and water.


For the past few years, our laboratory is working on developing novel biological control strategies for the management of invasive exotic plant species in South Florida. In another novel project, we hypothesized to mitigate freshwater toxins as well as marine toxins by applying microbiological techniques. Our group also involved in biological energy production such as biodiesel and hydrogen through plants and microorganisms. His old expertise on arbuscular mycorrhizal fungi and plant growth promoting rhizobacteria, their interactions with plants, and significant role in conservation/restoration measures for natural and agricultural systems continues.



Process Control By K Krishnaswamy Pdf 27

DOWNLOAD https://t.co/ece6mOUhSA 






where x, y are data points, εk(x), εk(y) are the distance from x, y to their k-th nearest neighbors, respectively, and α is a parameter that controls the decay rate (that is, heaviness of the tails) of the kernel. This construction generalizes the popular Gaussian kernel, which is typically used in manifold learning but also has some disadvantages alleviated by the α-decaying kernel, as explained in ref.3.


As this transformation is unitary, the inverse graph Fourier transform (IGFT) is f=Ψf^. Although the graph setting presents a new set of challenges for signal processing, many classical signal processing notions, such as filterbanks and wavelets, have been extended to graphs using the GFT. We use the GFT to process, analyze and cluster experimental signals from single-cell data using a novel graph filter construction and a new harmonic clustering method.


During the exploration process of oil wells, drilling operations are a critical stage and, due to its complexity, require high investments. This step involves the weight application on a rotating drill string, which causes the destruction of the geological formation linking the surface to the oil reservoir. The use of a drilling fluid is necessary mostly to chill the drill system, maintain pressure between wellbore and rock formation and remove drill cuttings from the bottom of the well, among others functions (Bourgoyne et al. 1991).


Drilling fluids are commonly heavy, opaque and abrasive. Besides that, drilling fluids have complex rheological behavior and tend to cause most online sensors to fail due to the high concentration of solids suspended (Caenn and Chillingar 1996; Magalhães et al. 2014). Therefore, monitoring and controlling operational conditions on such fluid are not trivial (Broussard et al. 2010). If a failure in hydraulic pressure control occurs, serious operational problems such as loss circulation, inefficient well cleaning or even a blowout may arise (Gandelman et al. 2013).


This paper demonstrates the preliminary results of a developed software which in a brief future could attend to an urgent need of drilling fluid operational condition control (Godhavn et al. 2011), such as wellbore pressure control in ultra-deep oil wells, or flow rate during gas invasion. This software was conceived to provide early diagnostics and interventions, which is, in this scenario, of most importance to avoid costly and even deadly drilling disasters (Oort and Brady 2011). Knowing that in drilling scenarios human resource can be expensive to maintain full time (Miller et al. 2011), the software primary objective is to automate every step needed to implement the controllers. The software was divided into three self-designed algorithms, named as automatic plant identification (API), auto-tuning (ABAP) and controllers auto-switch (CAS).


Even with optimum tuned controllers, unknown disturbances or major changes in the process may occur. This may lead to a necessity of re-adjustment in the previous calculated parameters. The auto-tuning feature (self-gain schedule) is capable of an online parameters recalculation, avoiding the necessity of the entire system re-evaluation. This is important for drilling fluids controlling mostly because of their diversity with distinct physicochemical properties. If auto-tuning is off, a controller once tuned for water-based mud may fail to an oil-based mud, in example.


The three modules presented were developed from reported experiences in the literature where modifications on classic PID controllers contributed to an overall improvement in process control. The main reason to merge classic control with techniques based on heuristic rules, such as Boolean and fuzzy logic, is to minimize inefficiencies which are singular to that specific process. Therefore, this work effort was to demonstrate that petroleum and gas industries may benefit from the techniques presented here, validated from experimental data, since the literature lacks on specific studies involving drilling fluids control characteristics.






The code first step is the selection of the manual or automatic mode, in other words, choosing to turn on the application or not. If manual mode is selected, the manipulated variable will be controlled manually by the user, configuring an open-loop system. If automatic mode is selected, a series of requests will be demanded by the application in order to allow the automatic control, configuring a closed-loop system.


In case of failure during search of the controller parameters, the user may proceed to identify the physical unit. This happens when the application is being installed for the first time on the computer or when the user wants to reevaluate the previously recorded parameter due to process changes. This feature was developed not only to minimize the time lost during manual calculation, but also to dismiss the presence of dedicated engineers making it accessible to lower levels of qualified manpower, which are generally at the front line of the process (Miller et al. 2011).


If the system is linear, one set of parameters is enough to control the entire range of operation. In case of nonlinearity, a 5-step calculus is made dividing the unit into 5 different parts, as shown in Table 2. Therefore, a nonlinear unit will have 5 different sets of parameters to control the entire range of operation. However, to avoid the constant parameters swap, the program uses an average weighted by the magnitude of each step.


The algorithm was developed to search for three types of inefficiency: slow offset elimination, overshooting (or dumping) and oscillatory behavior. For slow offset elimination, the application automatically increases Kc, which increases the controller gain as well as its integral action. As a side effect, the system may become unstable with high overshooting (dumping) effects or oscillation. In case that happens, the action to take is to decrease Kc. This cycle of attempts is maintained until three recordings are made. This recording procedure prevents the system to get stuck in an endless loop. After the third record, the application freezes the Kc value and starts tuning the integral parameter.


The range of actuation informs to application the tolerance to start changing parameters. For example, if tolerance is chosen as 5%, although the auto-tuning is on, if the controlled variable is distant from set point in 5% or less, the parameters will not be changed. This is fundamental as no industrial processes are capable of maintaining the controlled variable with an offset of 0%.


When the offset is minimized, the controlled variable may vary around the set point due to sensor or/and process noise. The oscillation frequency tolerance was created to distinguish between instability and natural oscillation. Usually, oscillation has a well-defined frequency, causing the controlled variable to fluctuate around set point in similar time intervals. Noises do not present such a mappable behavior.


The main HMI is loaded on the screen when the application is executed. From it, the user is able to perform the unit identification, turning on and off the automatic control as well as the auto-tuning feature. The CAS feature is configurable only by programming. If more detailed information is required, the user may load the second HMI, which contains more technical information and the whole system algorithm. The relation between both HMI is a relation MASTER/SLAVE, where the information entered by the user in the main HMI is passed down to the secondary HMI, which performs all the tests and calculations, returning them to the main HMI. The secondary HMI image can be observed in Fig. 4.


When the developed software is freshly installed in Windows environment, no unit and controller parameters are recorded in database. Therefore, the only action possible is to use the API feature. Figure 6 demonstrates how the application commanded the pump engine and how the flow rate was recorded while API was in action.


The five plant parameters (Delta, td, tau, t1 and t2) were reported for each step done. Also, all the combinations between the control strategy and its tuning method can be observed: P-CC, P-ZN, PI-CC, PI-ZN, PID-CC and PID-ZN.


At the bottom of Fig. 7, the weighted averages used to tune the controllers were presented. The dimmed region would be used only if the unit dynamic was linear. The whole process took no more than 16 min to be completed.


It can be observed in Fig. 8 that the controller with CC parameters (in blue) was faster when compared to ZN parameters (in black), when set point variation (in red) was imposed. This was expected because integral action is greater in CC parameters than in ZN parameters. Similar results were observed experimentally by GirirajKumar et al. (2010).

 0aad45d008

Back to comp.lang.basic.visual.misc | Previous | Next | Find similar | Unroll thread

Thread

Process Control By K Krishnaswamy Pdf 27 Esperanza Santrizos <santrizose@gmail.com> - 2023-12-25 22:52 -0800

csiph-web