01 June 2004
Correct instrumentation issues and educate operators before installing advanced process control.
As the chemical industry matures, fewer grassroots chemical manufacturing facilities are rising in North America today.
As a result, production increases and new product development opportunities target existing manufacturing facilities.
There are two options that a planner typically considers in determining how to make more product and to make it better and faster. The first option primarily addresses increasing throughput and involves de-bottleneck projects or expansion projects. This route typically requires a substantial capital expenditure and is mechanically based, requiring physical plant modifications.
The second option is to improve the efficiency of existing operations in the facility. This option addresses a combination of increasing production throughput and improving product quality. This option typically involves capital expenditure as well, but most often costs substantially less than the mechanically focused projects.
Optimization opportunities largely exist in the following areas:
- existing instrumentation and control system performance
- additional instrumentation and analysis capabilities
- operations training and procedural improvements
- advanced regulatory control enhancements
- advanced process control
The struggle for many companies is defining and then prioritizing the opportunities. There are multiple factors to consider in evaluating how to spend capital funds. The most obvious is to determine the potential return on investment (ROI) of all opportunities. However, evaluating the ROI of each of the opportunities on a standalone basis does not necessarily provide the best answer. One should consider additional aspects and then rank and weight the opportunities based on key decision criteria.
No single global answer is the right solution for all situations. Each facility should undergo evaluation on a case-by-case basis. An effective way of determining how to optimize a plant is to perform a plant operational review and develop a master plan. In-house resources, vendors, or system integration companies with the required expertise can perform these relatively low-cost reviews. The deliverable should be a well-documented master plan that identifies, prioritizes, and quantifies the expected benefits of optimization opportunities. In addition, it is vital to establish an existing baseline and define good metrics as part of the evaluation, so one can quantify improvements over time.
Many facilities, until recent years, initially installed instrumentation, valves, and control systems and only invested maintenance time and maintenance dollars on those installed end devices when they were "broke" and needed replacement. Instrumentation is at the very foundation of a plant's ability to control properly and achieve production goals. For example, if a valve indicates 100% open in the distributed control system (DCS), but it is actually 60% open in the field, either production throughput or product quality is suffering.
Over the past decade, preventative maintenance programs have become much more prevalent. These programs place key instrumentation on tickler lists and require more frequent and, many times, more in-depth instrument and equipment checkout. However, many of these programs still fall short, because they do not link the instrument performance to the process performance. Adding to this problem is the reduced process engineering staffs that many facilities employ today. A regular and thorough review of process data can identify instrumentation issues affecting plant performance.
With the challenge of reduced staffs, many facilities rely on the operations department to catch any issues and further diagnose the root cause of those issues before communicating the problem to the maintenance department. On obvious operational issues, this methodology may work. However, this organizational approach does not necessarily identify nor correct the more subtle process issues.
There are several methods to help plants identify instrumentation issues. The first involves new instrumentation diagnostics capabilities. Most instrumentation today is available with "smart" diagnostic features. This information can be tremendously helpful in analyzing subtle instrumentation performance degradation. Another method that works is to use process-modeling tools. There are tools available that perform instrument validation and data reconciliation calculations, which can be useful in diagnosing instrument performance issues.
In addition to instrumentation, controller performance also requires good tuning. If valves or instrumentation is not the same as it was when the facility did the original tuning constants derivation, then the tuning constants may need revising and/or updating. Tuning is a function of many things, and considering the entire process is important for optimal tuning. All major control loops require a regular performance review to determine if controller degradation has occurred, in which case tuning changes may be required. Online tuning tools can assist in keeping tuning up to date with process and instrumentation changes.
Although poor instrumentation performance can be problematic, many plants simply have inadequate instrumentation and online analysis measurements. There is often not enough information in the control systems to develop adequate strategies to control complex processes.
One example is that many facilities still rely on manual data logging by operations personnel. The logged information, from field gauges or readouts, feeds back to a control room operator who adjusts the controls. This forced approach of "living with what we have" rather than installing proper and necessary instrumentation and making the data available to the control system can cost companies substantial profitability in lost opportunities. A good rule of thumb is that if the information is part of the operator's decision criteria, then the information should be part of the control scheme.
A specific subset of instrumentation where a huge opportunity exists in many facilities is online analyzers. Unfortunately, in many plants, analyzers must overcome the reputation of being expensive and difficult to maintain. The availability of new technologies has continued to improve analyzer ease of maintenance and cost. Analyzers can provide valuable and timely feedback that easily pays for the expense of the analyzer.
Many times, operations personnel still take samples, run them to a lab for testing, and feed the results back to operations, typically via a lab information management system. The delay between the sample and the results depends upon the testing requirements, but can be substantial. This delay can mean hours of off-spec product. With ever improving technologies, online analyzer systems—if properly designed, installed, and maintained—can be a great method of substantially reducing, and in some cases eliminating, this feedback cycle time delay. Having this feedback automatically built into the control, or even giving an operator the information faster, is an opportunity to make timely control adjustments and save hours of off-spec product.
One of the most difficult challenges that chemical plants face is shift-to-shift operational variability. Chemical companies spend large amounts of money each year updating procedures and providing new and refresher training to operations personnel. However, the human factor continues to affect the job. Especially because the plant operations groups are typically not immune to staff reductions, they often face the challenge of conducting many activities during a shift. The impact of the difference in human perspectives, ideas, and interpretations is visible in both product quality and production throughput on virtually a daily basis in many plants. This inconsistency is a critical driver causing companies to invest in additional automation. Automation reduces human decisions points, which can be the source of substantial production inconsistencies.
As the role of automation in plant operations grows, the role of control room operators changes. Without automation, operators spent a significant amount of time making control moves and adjustments. With automation, the operator primarily monitors performance and looks for opportunities to maximize plant performance. Training programs have changed to focus more on abnormal situation management rather than on normal daily control. With the elimination of the need for frequent moves by operators, many feel that operators have lost some of their process understanding. However, experts say that process understanding is even more critical to operators today as they learn to find the opportunities to push the plant toward maximum potential. This will not happen if operators do not have a strong fundamental understanding of the process, of the controls, and of how they link to the overall plant economics.
Training simulation packages are becoming a more predominant part of the approach of many chemical companies. Process simulation packages can help operators better understand the process and the cause-effect relationships of control moves. The investment in training simulation packages can easily be justified by avoiding just one significant process upset caused by operator error.
Linking the economics with operator actions is one of the growing industry trends. Key performance indicators (KPIs) are becoming readily available to operators through historian systems. These KPIs actually provide critical production and economic metrics to the operator such that the decision process can become more business driven.
Advanced regulatory control
Two extreme philosophies often lead to less than optimal control performance. The first is that the proportional, integral, derivative (PID) algorithm is the only control available. There are companies that have never used or considered using any strategies outside of the PID algorithm. These plants, which are typically the smaller chemical companies with small or no engineering staffs, usually have one or two constant control problems that are always a source of operational upsets and turmoil. The second approach is any problem not solved with PID requires ad-vanced process control (APC), which many consider synonymous with multivariable control. These facilities are often ready to spend large amounts of money on multivariable control products that are not necessarily required.
Many available options fall between these two extremes. A few common advanced regulatory control strategies to address more complex control problems are cascade control, feedforward control, ratio control, and pressure and temperature compensated control.
Normally, one can install these algorithms at the base-level control system at relatively low cost. The benefit is the plant gains a more sophisticated control strategy to address the more complex problems without a large investment or overly complex scheme. A good example of a loop most often requiring advanced regulatory control is pH. The PID control algorithm is usually not effective in pH control due to the nonlinearity issues.
A pH control system can be a relatively simple on/off control loop in some batch processes. At the other extreme, it can be a complex, feedforward/feedback system with multiple sequenced valves for continuous neutralization over a wide, dynamic pH range.
Basic PID control should serve whenever possible. In control, simpler is better, and overcomplicated controls frustrate operators. However, for situations where basic PID is not adequate, these advanced regulatory strategies are a feasible solution to improve control performance without more sophisticated and higher priced packaged software solutions.
Advanced process control
There are complex control problems out there that can only reach resolution through APC solutions. The decision is not necessarily an "either/or" with the previously discussed advanced regulatory control strategies. Oftentimes advanced regulatory control strategies are an effective, and at times, necessary building block for advanced control and optimization programs. However, the basic operation of the instrumentation, valves, and base PID controllers is absolutely required for adequate performance of installed APC.
Advanced process control and optimization is comprised of many components including, but not limited to, multivariable control, process modeling, inferential predictive models, transition control, and real-time optimizers.
Determining which components are required can itself be a challenge. Each of these has benefits that address specific problems. Here they are with particular applications for each.
Provide a matrix dynamic
Multivariable processes are those where changes to the set point of one variable affect the process value of multiple other variables. These processes generally require multiple operator moves to achieve a single objective.
Most of the multivariable control tools available provide a matrix of the dynamic models of each of the key manipulated and controlled variables. These relationships operate from models that stem from planned step-test analysis, process simulation, or historical data analysis. Multivariable controllers operate against constraints and typically maximize throughput production within the boundaries of quality control. Multivariable control may be useful when:
- The variable relationships of a process require a complex series of operator control moves.
- After implementing PID control, the process does not reach the desired degree of control.
- The nonlinearity of the process complicates the control picture.
Process modeling captures the dynamics of a process. For complex processes where the understanding of reactions is limited, models can provide a great insight into the process. One can distribute this information to the historian system for review by the operators or link it to some decision criteria within the control system.
Inferential predictive models work in place of analyzers, or in place of lab results to predict compositions, product properties, or other key information. Predictive models can be empirical in nature or be neural networks or process models. One can incorporate model results into the control solution, or simply provide them to operators as an additional data point in the decision-making process.
Many chemical companies make multiple grades of products. Product losses due to transitions can be large. Transition control programs that reside on the base-level control system are available. These programs help to manage transitions and minimize transition time from the on-specification production of one product to the on-specification production of another.
Real-time optimizers are usually the highest level of control and are able to determine optimal operating points within specified boundaries to meet predetermined goals and objectives.
Many of the optimization benefits can take place through opportunities other than APC. However, selective use of various APC applications is generally the best way to get the remaining benefits that are not achievable through other, more traditional methods. The challenge of APC is the expertise required to install and maintain it. However, outside resources can handle this issue if qualified in-house resources are not available. The ROI on APC projects when implemented at the right time and in the right manner is usually high.
Many facilities get stuck
There are likely many opportunities to consider for potential plant optimizations. The assessment of these opportunities and determination of how to proceed is oftentimes where facilities get stuck. The facilities have confrontations and emotional debates about what the right next step is, and many times the next step never occurs.
There is a methodical approach to addressing, prioritizing, and developing action plans for plant optimization opportunities.
The first step in the approach is to define the key criteria for the assessment of opportunities and to weight or prioritize these criteria. As each opportunity becomes clear, one compares and scores the opportunity against the criteria. Using the prioritization as a guide, develop a master plan or road map to outline the sequenced implementation of the opportunities. The final key component prior to implementation is the development and documentation of metrics. Engineering must establish the metrics and measure the implemented projects against them to determine project success.
The method of developing the criteria for evaluating opportunities can be very systematic. There are generally three categories of problems—fixed input, fixed output, and neither fixed input nor fixed output—with associated economic criteria.
Understanding which category of problem a plant is facing is critical to developing the correct solution. In general, most facilities look at the category where neither the input nor output is fixed. This is the case because most companies are willing to spend money, always within limits, if the return provides an acceptable level of benefits.
Most companies have spreadsheets or economic programs to calculate an ROI, rate of return (ROR), or payback period. These calculations provide a numerical representation of the expected benefits versus costs over some time period. There is typically a minimum threshold required for a project to receive consideration for capital expenditure. Assuming there are numerous opportunities that meet this minimum financial criterion, then the difficulty becomes how to compare and prioritize the various opportunities.
An effective method for comparison is a prioritization criteria matrix, which is simply a weighted list of the criteria important to a plant. This matrix assigns a quantifiable score to each opportunity.
The benefit of this process is that it requires decision makers to agree on the grading criteria and the relative importance of each factor. This method does not completely eliminate subjectivity. However, it does document the decision process. Having a documented, quantifiable score helps one more easily compare opportunities against one another.
Master implementation plan
After deciding which opportunities to pursue, it is important to have a master implementation plan. A good master plan outlines a systematic approach to implementing multiple opportunities in a sequential manner to maximize benefits in a minimal amount of time.
The plan should:
- not only address the immediate needs, but also provide a three to five year look ahead
- account for anticipated process, personnel, and market changes as much as possible
- ensure that any prerequisite steps are complete before implementation of each opportunity
- provide an implementation schedule for activities
- address specific economic goals and metrics
The importance of having a documented approach to taking advantage of available opportunities cannot be overemphasized. When a plan exits the creative process and broadcasts to the company proper, it not only allows a company to better develop capital budgets and to utilize available funds, but it also encourages engineers and operators to think in terms of the future. For example, if a company is installing a small valve and instrument project, an engineer may order smart technology if the documented plan is to begin using those diagnostic capabilities for preventative maintenance in the next two years.
However, if no plan has been developed or communicated, the likely approach is to purchase and install instrumentation consistent with the instrumentation already installed, which potentially does not take advantage of smart capability.
This process is benchmarking
One of the most significant challenges for optimization projects is measuring benefits. This often translates into difficulty getting capital funding to execute optimization-related jobs. When benefits are obvious but hard to quantify many managers and many companies are not willing to take a chance of having to justify funding.
Several factors make the metrics for optimization projects difficult. Optimization projects tend to focus on larger areas of the process and address quality or throughput issues that have many factors. At the same time that optimization projects are in progress, other projects may also be tackling the same issues, and so the benefits credit to multiple jobs. In addition, other factors can often affect product quality variation such that even if an optimization effort is successful, the end results may not be evident at times.
An absolute requirement of moving forward with optimization projects is to establish a baseline before beginning implementation. Historian systems are usually the best manner of building metrics. KPI screens can be developed on the historian that calculate and illustrate the current performance of a given area, unit, or plant. The goal of KPIs is to link the specific actions of operations with the overall economics of the plant in either a numerical or visual way. If an installed historian system already exists, then KPI implementation can be relatively easy. The KPIs should establish and document the current plant baseline performance.
In addition to understanding the current performance, one should compare this performance to similar processes, to determine the improvement potential. This process is benchmarking.
Examples of benchmarking using manual data or industry performance figures collected annually are common, e.g., the Solomon study in the oil-refining sector or the Townsend report for polymers.
Before implementing any given optimization opportunity, establish a written set of economic goals. Whenever possible, these goals should be in terms of existing KPIs to simplify measurement. However, it is important to develop these goals in concert with technical resources and management so that project expectations are clear. A project's implementation goals should also include a realistic duration of time before success is measured.
To be a secondary activity
In most chemical plants, optimization opportunities exist that either have not been identified or have not been implemented. These opportunities can provide significant improvement to bottom-line profits while typically requiring relatively low capital investment. The first step is an initial investment to identify optimization opportunities, prioritize them, and develop an implementation plan.
Plant optimizations involve every aspect of the plant including maintenance, engineering, operations, and management. All plant departments should focus on maximization of profitability and product quality, while maintaining a safe and environmentally friendly work environment. Using the systematic methodology outlined, one can do a quantifiable comparative analysis to determine which opportunities should be pursued, and in what order.
Today most companies face the challenge of reduced staffs that mainly focus on daily plant operations, rather than optimization. However, evaluating plant optimization cannot continue to be a secondary activity. As the chemical industry market continues to grow globally, taking advantage of optimization opportunities will be a differentiation between chemical companies that operate profitably and those that do not.
Behind the byline
Dan Roessler has a degree in electrical engineering. He is a manager at Mangan, Inc., in Texas, which performs control and optimization studies and projects in the oil, gas, chemical, and refining industries. Write him at email@example.com .
Initialisms and concepts
Advance process control (APC) process control strategies beyond straightforward PID loop control, which experts usually call "classical" advanced control. These advanced strategies involve a combination of PID loops, dead-time compensators, lead/lag feedforward function blocks, and single-variable constraint controllers.
Cascade control control action in which the output for one controller is the set point for another controller (ANSI/ISA-77.42.01-1999)
DCS distributed control system
Feedforward control control action that compensates for the effect of a sensed input or disturbance. Engineers also refer to this as open-loop control or anticipatory control.
Key performance indicator (KPI) a quantifiable measurement agreed to beforehand, which reflects the critical success factors of a process or organization
PID control proportional (P) plus integral (I) plus derivative (D) control is control action where the controller output is proportional to the error (P), its time history (I), and the rate at which the error is changing (D). The error is the difference between the observed and desired values of the variable that is under control action.
Ratio control control action that maintains a predetermined ratio between two or more variables in a process
ROI return on investment
Return to Previous Page