Enabling Advanced Automation for Pipeline Operations - Part II

January 24, 2023

Author: Jayson McAllister, M.Sc., P.Eng., Kamaal Mahmud, P.Eng.

Introduction
With reduced spending on new infrastructure, pipeline operators must take advantage of new technology to optimize existing assets and continue to improve their bottom lines. Emerson’s DeltaV DCS can be an exceptional tool in developing optimization control strategies and cross-PLC applications for pipeline operators. DeltaV can interface with any existing SCADA system via OPC communication, facilitating easy implementation and maintenance of advanced automation solutions. Advanced automation is essential for reducing operator workloads, improving process reliability, pushing process constraints to maximize profits, and coordinating siloed assets. Part II of this blog will focus on advanced automation applications that have been implemented by Spartan Controls for a large pipeline operator customer, made possible with Emerson’s DeltaV DCS.

Butane Blending Model Predictive Control (MPC)
Pipeline operators often source crude from dozens of different producers. The compositions of these crudes are subject to relatively high variability. When negotiating contracts between pipeline operators and downstream clients, the crude oil supplied must meet certain specifications that are agreed upon by both parties. Crude oil is typically sold by volume. The high composition variability of these products generally means that the specifications allow for a relatively large range of product compositions. Pipeline operators can use these specifications to their advantage by increasing the sales volume of the crude by blending in lower cost components. This may include blending certain types of hydrocarbon components such as butane (C4).
 
Butane can be blended into crude oil to increase the volume of the crude oil for sale. By adding more butane, more revenue is generated due to the significant cost differential between butane and crude oil. Specifications for custody transfer of crude oil often use density and Reid Vapour Pressure (RVP) to regulate product quality. Any product outside of the defined limits results in a violation of the contracts and the potential for monetary punishment for the pipeline operator. When blending in butane, the pipeline operator must ensure that the low density and high RVP constraints are not violated. RVP measurements are taken by expensive analyzers, which generally have large dead times in between sampling times, which can exceed 8-10 minutes. Large dead times mean that changes made now are not observed in the RVP measurement for that length of time. The CRO is typically responsible for micromanaging the butane flow input manually to ensure constraints are met. This presents a difficult and time-consuming task for the CRO since traditional PID control is not well suited for large dead time control applications. To overcome the large dead time challenge as well as optimizing for multiple constraints, Spartan Controls developed and deployed a Model Predictive Control application for blending utilizing the embedded function MPC function block in the DeltaV DCS.
 
The SCADA system collects process data from the field PLCs and equipment and communicates this to DeltaV via OPC. Figure 1 describes the communication network architecture of the overall system. SCADA collects the relevant measurements from the PLCs in the field and communicates them via an OPC Read pipe. DeltaV receives this information using a customized function block emulation layer developed by Spartan Controls. The emulated modules are then readily available to facilitate advanced control applications. These applications can then communicate the required control actions back through the emulation layer, through an OPC Write pipe, and back to SCADA, where they can be sent to the relevant local PLCs for implementation.

Figure 2 DeltaV-SCADA Communication Interface
Figure 2 DeltaV-SCADA Communication Interface
Emerson’s DeltaV Predict Pro consists of a MPC with an embedded LP (linear program) optimizer. The process model gains and the optimizer are used to determine the steady-state targets for the density and RVP measurements. In this case, the optimizer is set up with specific objectives: maximize the RVP while adhering to the minimum density constraint. These targets are then given to the model predictive controller to push these variables to their targets. MPC uses an internal model of the system to predict where a process is trending based on its past measurements and optimizes the future trajectory of the manipulated variable (Butane Ratio Set Point) which will drive the system to its optimal targets. MPC excels in applications for constraint handling, large dead times, and/or multivariable systems.

This application was able to significantly reduce operator workload for the customer (on average 75 to 100 fewer operator changes per day). It also reduced RVP target error mean and variability by 98% and 34% respectively, while increasing the amount of blended butane volume by 5.6% in its first month of operation! See Figure 3 below.
Figure 3 Butane Blending MPC Application - RVP Target Error Histogram
Figure 3 Butane Blending MPC Application - RVP Target Error Histogram
Procedure Automation for Pipeline Operations
Procedure automation should ideally exist at the local controller level (typically within PLCs) whenever possible due to the potential for communication errors between SCADA and the local controllers. Pipeline operators have generally implemented excellent use-cases of procedure automation that occur on the PLC-level. PLC-level procedure automation includes simple logic such as opening valves when boosters or pump start requests are initiated. However, automation can be lacking when it comes to cross-PLC applications. Both DeltaV and SCADA can provide automation over multiple PLCs. SCADA uses script-based logic to implement advanced logic, which is time-consuming and cumbersome to create and maintain, and it consumes a large amount of CPU processing power on the SCADA system. As a result, more complex cross-PLC procedure automation should exist in a DCS system designed for it, utilizing SCADA’s ability to see and interact with equipment spread across multiple PLCs. DeltaV DCS has built-in capabilities and functions that are well suited to implement procedure automation of varying complexity, which is also easily maintained. One of the major applications for cross-PLC applications is giving the Control Room Operator (CRO) the ability to operate a pipeline at the click of a button.
 
Spartan Controls implemented procedure automation on a DeltaV DCS for a 12” pipeline that transports batched products. This automation allows the operator to automate various manual processes in the operation of the pipeline. The pipeline undergoes multiple starts, stops, and product switches between C2 and C3 throughout the day and requires the operator’s full attention to perform these operations safely and in a timely manner. Multiple automation sequences were designed to provide the following automation functionality for the line:
  • Start line on C2
  • Start line on C3
  • Switch line to C2 operation
  • Switch line to C3 operation
  • Inter-Tank Transfer at the initiating station
  • Line shutdown
The DeltaV HMI is shown in Figure 2. The console operator uses the SCADA HMI to select the desired mode of operation and which assets to use, and DeltaV automatically reaches the desired state. The functionality is achieved using four sequences: one overall main sequence to handle common assets, and two sub-sequences for C2 and C3 operation that will coordinate with the main sequence. The inter-tank transfer is handled by a stand-alone sequence.
Figure 1 Pipeline Procedure Automation DeltaV HMI
Figure 1 Pipeline Procedure Automation DeltaV HMI
As a result of this automation, the operators were able to save 1h per day in total time spent on the operation of this pipeline. The reliability and consistency of the application helped to reduce start-up times, reduce failed start-ups, and improve the safe operation of the pipeline.

Reid Vapour Pressure Kalman Filter

The process industry routinely encounters important but difficult-to-measure process variables. The Reid Vapour Pressure (RVP) is an important measurement that can be used in blending applications to push RVP constraint limits and maximize profits, as mentioned earlier in the Butane Blending MPC section. Applications that rely on this measurement for closed-loop control are only feasible as long as the RVP measurement is accurate, reliable, and continuous. RVP is a challenging measurement as the analyzers used to measure it need time to process the samples. This usually results in only a handful of measurements available every hour. In addition to this long sampling time, these analyzers on occasion give unreliable measurements due to sampling system errors or analyzer faults. These data anomalies can lead to instability and over/under-blending when in closed-loop control.
 
An array of machine learning methods are available for use in order to predict measurements. One such method, which was implemented by Spartan Controls, is the Kalman Filter. The algorithm uses a series of analyzer measurements over time and a discrete-time state-space process model to produce more accurate estimates of a particular variable of interest. This Kalman Filter was used to improve an RVP analyzer measurement and was able to be fully implemented within the DeltaV DCS.
 
The implemented Kalman Filter algorithm attenuates the effect of measurement noise and provides a faster and continuous estimate of the true RVP in between the analyzer sampling times. Figure 4 shows an observed RVP measurement in blue and the Kalman Filtered RVP in green. The figure also shows how the RVP can be predicted when no measurement is available, as well as reject unexplained measurement spikes.
Figure 4 Representation of the Behaviour of the Kalman Filter
Figure 4 Representation of the Behaviour of the Kalman Filter
The combination of Kalman Filtered RVP with the Butane Blending MPC yielded further process improvements:
  • 59% reduction in over-blending events (i.e., RVP greater than product spec limit)
  • 34% reduction in RVP target mean error
  • 17% reduction in RVP target variability
  • 39% reduction in Butane blend ratio mean
The faster and more reliable RVP measurement allowed the control engineers to confidently tune the MPC more aggressively, resulting in improved control performance. The reduction in process variability and over-blending events supported pushing the operating RVP target 17% closer to the hard spec limit, resulting in further revenue increase due to increased butane blending.
 

Automated Asset Handling Upon Line Scraper Arrival

Line Scraping is a necessary task in the pipeline industry. Line Scrapers are used to clean and inspect pipelines to ensure their continued safety and reliability. Scrapers are launched directly into an operating line and travel through the pipeline scraping the inside to remove debris. When line scrapers approach pump stations and other infrastructure, flow path, line flow rates, and/or pumps need to be adjusted to allow the line scraper to pass. Spartan Controls implemented an advanced automation application in DeltaV that tracks the line scraper as it travels down a pipeline and automatically arranges assets to allow it to pass unencumbered. This position estimation soft sensor uses line fills between stations, volumetric flow rates at various injection/withdrawal points, and intelligent use of line scraper position switches to estimate the position of the scraper at any given time. The algorithm can assess estimation bias and learn to correct it over time, improving accuracy under changing conditions (i.e. crude compositions, temperature, pressure, flow meter calibrations).
 
This position estimate was used to estimate the arrival time of the line scraper at different assets and perform several types of actions:
  • Provide the control room operators (CROs) advanced warning of scraper arrival and prompt them to complete certain actions manually
  • Automatically open and close valves to allow line scrapers to pass by infrastructure
  • Automatically disable blending applications and isolate blend loops to allow the line scraper to pass by, and then restart blending applications afterwards
The application has resulted in more consistent operation of the line with the scraper. On average, it has reduced the amount of time the blending operation is shut down from 1h to 30 minutes, resulting in an additional 26 hours of butane blending over a year.
 

Closing

Implementing advanced automation applications, such as the ones mentioned above, will position pipeline operators at the forefront of a technology-driven economic revolution. Targeted and relevant overarching control and optimization has the potential to unlock significant process improvements for pipeline operators by providing safe, reliable, repeatable, and consistent operations across all assets. DeltaV DCS and Spartan Controls have the ability and expertise to drive real-world results that are sustainable over time. DeltaV has the built-in functionality to interface with any existing SCADA system and tools to implement basic and advanced process control applications. In addition, it enables the incorporation of machine learning, advanced optimization, and many other state-of-the-art control solutions, which are not traditionally deployed in the pipeline industry. At Spartan Controls, we invite you to “Experience Industrial Innovation”.
 
Learn more about our Process Controls https://www.spartancontrols.com/solutions/process-control/
 
Advanced Process Control Engineer
Jayson McAllister
Advanced Process Control Engineer
Advanced Process Control Engineer
Kamaal Mahmud
Advanced Process Control Engineer

Related Stories

Our website uses cookies. By continuing to browse our site you are agreeing to our Cookie Policy.

Back To Top