Professional Documents
Culture Documents
EPRI 3412 Hillview Avenue, Palo Alto, California 94304 PO Box 10412, Palo Alto, California 94303 USA
800.313.3774 650.855.2121 askepri@epri.com www.epri.com
ORDERING INFORMATION
Requests for copies of this package should be directed to the EPRI Distribution Center, 207 Coggins Drive, P.O. Box
23205, Pleasant Hill, CA 94523, (925) 934-4212.
Electric Power Research Institute and EPRI are registered service marks of the Electric Power Research Institute, Inc.
EPRI. POWERING PROGRESS is a service mark of the Electric Power Research Institute, Inc.
Copyright 1999 Electric Power Research Institute, Inc. All rights reserved.
CITATIONS
This report was prepared by
Raytheon Engineers & Constructors, Inc.
100 Corporate Parkway
Birmingham, AL
Principal Investigator
J. M. Mendel
This report describes research sponsored by EPRI.
The report is a corporate document that should be cited in the literature in the following
manner:
Survey of Instrumentation and Control Practices in the Process Industries for Application to the
Power Utilities; EPRI, Palo Alto, CA: 1999. TR-112230.
iii
REPORT SUMMARY
practices that have proven beneficial in process industries and assess their usefulness to
fossil generating stations.
Results
The report is a compendium of responses to survey questions in four areas: Planning
and Justification of Automation and Control System Upgrades; Digital Control System
Architecture; Digital Control System Vendors; and, Implementation Procedures. It
includes comments, discussion, conclusions, and recommendations from a panel of ten
I&C experts, each with more than twenty years experience upgrading process industry
systems. Interviewee responses are either paraphrased or given verbatim, as
appropriate. In some cases, interviewees ranked control system upgrade issues in order
of importance. This report will help utilities benefit from the experiences of process
industries as they modernize plants and adjust to a competitive environment. Based on
this timely information, utilities will be able to better prioritize, justify, and plan I&C
modernization efforts at their plants.
EPRI Perspective
The utility industry should pay careful attention to process industry approaches and
learn from their experiences. While utilities are only now learning to operate as
competitive businesses, process industries have been doing this for many years.
Presumably, their decisions to update their I&C and information management systems
have been motivated by the need to maintain a competitive edge. Cost justification of
I&C upgrades has proven particularly problematic for utilities. It should, therefore, be
instructive for the utility industry to study the technologies and upgrade approaches
that have been adopted by these traditionally competitive industries.
TR-112230
Keywords
Controls
Plant retrofits
Instrumentation and control
Automation
vi
ABSTRACT
To develop an understanding of modern instrumentation and control (I&C)
technologies that might benefit the utility industry, analysts interviewed ten I&C
experts with broad experience in upgrading control systems in the process industries.
The interviews were conducted to discover practices that have proven beneficial in
process industries and assess their usefulness to fossil generating stations. This report is
a compendium of those interview responses and is divided into four categories:
Planning and Justification of Automation and Control System Upgrades; Digital
Control System Architecture; Digital Control System Vendors; and, Implementation
Procedures. Information here will help utilities benefit from the experiences of process
industries as they modernize plants and adjust to a competitive environment. Based on
this timely data, utilities will be able to better prioritize, justify, and plan their I&C
modernization efforts. Cost justification of I&C upgrades has typically proven
problematic for utilities. It should, therefore, be instructive for the utility industry to
study the technologies and upgrade approaches that have been adopted by these
traditionally competitive industries.
vii
ACKNOWLEDGMENTS
The authors and EPRI would like to acknowledge the following experts in
instrumentation and control who made this study possible by contributing their time
and expertise as interviewees.
Mr. Robert W. Barber
We also thank the following people who reviewed drafts of the report and provided
comments and recommendations that were used to improve the final version.
Ms. Teresa Taylor
And finally, we would like to acknowledge the staff of the EPRI I&C Center, who
advised the author on the important fossil plant issues.
Mr. Duane Bozarth
Mr. Robert Frank
Mr. Cyrus Taft
ix
CONTENTS
INTRODUCTION............................................................................................................. 1-1
xi
CONCLUSIONS.............................................................................................................. 6-1
6.1 Planning and Justification of Digital Control System Upgrades...................................... 6-1
6.2 Digital Control System Architecture ............................................................................... 6-2
6.3 Digital Control System Specifications and Vendors ....................................................... 6-2
xii
SURVEY INTERVIEWEES..............................................................................................A-1
xiii
1
INTRODUCTION
Implementation Procedures
The interviewees for the survey were representative of those people who have dealt
with these issues. A listing of the interviewees, with their organizations, titles, and
experience is shown in Appendix A. Note that these people each have 20-30+ years
experience in digital control systems during the period when these systems were
displacing panel-mounted, analog, control systems in the process industries.
The following are the typographic conventions used in this document. Direct interview
comments, paraphrased by the investigator, have been italicized. In a few cases, where
interviewees responded in writing, the comments are italicized and shown within
double quotation marks ( ) to indicate that the comments have been repeated
verbatim. The investigators conclusions and recommendations, based on all of the
interviews, are shown in boldface.
1-1
2
PLANNING AND JUSTIFICATION OF DIGITAL
CONTROL SYSTEM UPGRADES
In the process industries beginning in the late 1970s, Distributed Control Systems
(DCSs) began replacing panel-mounted single loop controllers (pneumatic, analog
electronic, and digital electronic). In parallel with the regulatory instrumentation loop
logic shifting to DCSs, motor start/stop (interlock) logic, implemented with hardwired
relay panels, migrated either to Programmable Logic Controllers (PLCs) or to the DCS.
The cost/benefit analysis and justification of digital control systems was a popular topic
in process industry technical journals of the early to mid 1980s. Cost factors, such as:
controls equipment prices, wiring, floor space, control room/rack room HVAC,
operational manpower requirements, maintenance requirements, etc. were estimated
and compared between panel-mounted and DCS architectures. Process operational
incentives (benefits) due to such effects as: production variation reductions, raw
material usage reductions, control system reliability improvements, integration
capability with plant information systems, etc. were also estimated and compared
between architectures.
Although the digital technology has changed, the procedures for making decisions
about control renovations to processing plants has remained constant: As a first step,
management still insists on a feasibility study or cost/benefit analysis of a proposed
project. Not only must costs and benefits be estimated, but the proponents of the
renovations must understand the risks embodied in these estimates. What has been
learned in the process industries in the many repetitions of this exercise by the
interviewees? This section of the report will examine the following topics in this
category:
1. Initial Cost Factors Associated with a Digital Systems Upgrade,
2. Continuing Expense Factors Associated with Digital Systems,
3. General Categories of Benefits of Digital Control Systems,
4. Benefits of Specific Advanced Control Loop Logic Algorithms,
5. Benefits of Batch or Sequence Controls,
2-1
2-2
Significant but one of least difficult to estimate. Accurate I/O count will lead to an accurate
equipment cost.
Significant and highly dependent on getting an accurate I/O count and agreement on the
scope of the operator interface.
Costs are significant, but easy to define. Actual cost typically comes close to estimate.
Typically retrofit must install new equipment while old equipment is still in place and
running. Installation must be carefully planned to be able to remove old equipment after
new is started up, e.g. new cable should not be laid over the old.
Typically the largest impact. Cost of operating while constructing and demolishing old
equipment requires detailed planning, and is usually underestimated.
Rank No. 1(Tie with Preparation of Control Requirements Documentation for highest) in
analysis impact. Weve gotten burned on several problems over the years: We have had
problems getting the old equipment out after the new was installed. Grounding was usually
underestimated. The cost of air filtration to remove dust and air-lock double doors to the
2-3
control and rack rooms was underestimated. There was also the cost of providing backup
power; we have gone to emergency generators to hold power after battery backup units
expire.
Rank No. 2 (out of 8, 1 highest) in analysis impact. Primary problem is having to work in an
operating control room, where the operating organization may not be as dedicated to the cutover schedule as the engineering/construction organization. Often hidden problems will
come to light after the project has begun: e.g. lead paint on columns to be removed, new
lighting requirements, existing HVAC problems, etc.
This is generally underestimated, due to a lack of understanding, by the estimator, of the site
organization and the condition of the site equipment.
Significance depends on availability and validity of existing documents, e.g. may decide to
run entirely new field cables if documentation is poor. Typically the cost is not relatively
significant unless there is a supervisory system included, such as an IBM 1800.
Projects that go well will have DCS users taking an active role in this activity and
understanding exactly what they are asking for.
One interviewee disagreed with the above and stated -- Cost has low sensitivity to
estimation error, because this is a routine, straight forward, procedure for an upgrade.
However, the investigator suggests that the following two responses deserve attention
for their insight into the importance of this factor:
2-4
Rank No. 1 (Highest, Tie with Site Preparation) in analysis impact. We originally
performed a loop-by-loop swap to the DCS, and got a very low cost for this. In fact, this
effort should have had a much larger scope, with a larger cost, and it is a worthwhile
investment to get more out of the DCS.
Rank No. 1 (out of 8, 1 highest) in analysis impact. This takes a lot of time, particularly
from the people with the least amount of time available. Getting their input is important.
To get the maximum value from a DCS replacement, more than just the DCS
implementation of the same control loops is necessary. As we will emphasize later in
this report, when benefits are discussed, advanced control techniques will require
process review, analyses, model building, and model validation with process data.
Configuration typically not relatively significant, unless adding upper level or advanced
regulatory functions.
This can be easily subject to cost over-runs, particularly if the contractor offered a low bid
price and is losing money on the job.
However, one interviewee probably identified the root cause of the diverse response
with his comment: Easy to perform, once the logic has been identified and documented
If the logic and interface requirements are clearly defined in the control requirements
documentation, configuration will not be a cost or schedule issue.
Verification of the hardware at the vendors site is straightforward and not difficult to
estimate. Verification of the configuration is a lot more difficult and therefore difficult to
estimate. A test plan should be written. Many customers require simulation using loop back
to verify each loop.
2-5
This is a time consuming effort, but it is a must. This is the last chance to see how the
system will work and re-do.
We do a complete hardware staging using a simulator. We check the configuration and the
graphics at the staging. This pays by eliminating startup surprises.
A simulator is best for verification of logic, if the job can afford it.
One interviewee stated that staging was unnecessary -- Recommend deletion of this item.
Staging is unnecessary. A major oil refiner is doing zero staging.
Staging is costly, particularly if logic verification cannot be accomplished as a
previous, separate, step. It also requires a spot in a tight schedule where there is
temptation to shorten duration. It does eliminate startup surprises. An experienced
team, implementing well defined logic on a seasoned DCS architecture, could solve
any problem, that would be found at staging, in the field at startup. Typically,
however, staging is a good insurance policy.
This is usually a well defined problem. Sometimes there is problem communicating the
requirements to the construction company, leading to a cost estimation problem.
I/O checkout is not a serious cost factor; however, it should be defined in detail to the
construction contractor.
High fidelity simulation is not worthwhile if the operators know the process and are being
trained on the control system. Teach the process with a low-level simulator to keep costs of
simulator low and certain.
Cost has low sensitivity to typical estimation errors, once the number and types of people to
be trained is decided.
This has not been an analysis problem. We spend more with the DCS than we used to.
This is easy to define but difficult to implement. Operating managers will make the
commitment to training, on the dates that it is scheduled and available from the DCS vendor,
but if anything else comes up, they will tend to sacrifice training.
The importance of training is the most under-estimated factor in the long term success of a
project.
2.1.8 Startup
Comments on Startup as a cost factor were fairly well in agreement:
Not significant if the other steps are done correctly, particularly Verification and Staging. A
major chemical company will not do a startup until the logic has been simulated.
Not significant.
Costs here should be low, if everything listed above has been accomplished.
Ive been blessed with good startups. I attribute that to diligence to Verification and
Staging.
No problem.
2-7
The main purpose of the Verification and Staging step is to insure a trouble free
startup, and this will indeed be the case, if that preparatory step is diligently
executed. Job cost accounting systems should be capable of identifying the proper
source of cost variance, should meeting other objectives increase the control system
startup cost.
Continuing costs will depend on which of two typical strategies is chosen: Freeze system
architecture, or keep architecture up-to-date. Freezing will lead to low annual costs until
entire system is scrapped and replaced. Keeping up-to-date with software releases, new
console hardware, etc. may be expected to cost about 10% of the system cost per year.
I recommend freezing system architecture for life of system. If you dont need it, dont buy
it.
For a DCS, continuing hardware costs are minor, just the replacement of failed components.
Software upgrades and enhancements can be expected to cost 5% of the initial cost per year.
Keeping people up to date on system is important and can be expected to cost 2% per year.
Continuing costs should be low, because initial software should be used for several years
without upgrading. Training costs will depend on other factors, e.g. employee job changes.
Count on events such as software upgrades, Y2K, etc. We get a preliminary quote from the
Vendor and put that number into the plan. The 5% of the initial cost per year (suggested by
another interviewee) is probably pretty accurate.
We spend about $50-70,000 per year per control room/rack room to replace spares. We also
have the cost of software licensing and support. Our DCS vendor warehouses spares here, so
we only replace what we use. Training costs are definitely higher than with panel-mounted
instrumentation. We send each new person on DCS maintenance to the DCS vendors
school for 3 weeks.
2-8
Software upgrades for our DCS probably average 10% of the initial cost per year. 2% of
initial cost per year for training is probably about right. We have a problem identifying the
right people to train. In several cases the people trained do not develop the overall diagnostic
skills and motivation necessary to make a good DCS technician.
Software and hardware maintenance should be 3-5% of initial cost, if the right DCS vendor
is selected. Some people do incur continuing DCS vendor costs greater than 10%.
To get the full benefits of the digital system upgrade, assume in the analysis that the
system will be kept up to date at an annual cost of 5-10% of the initial cost per year.
Also include 2% of the initial cost per year for training.
2.3.1 Smaller Process Variation and Operating Closer to the Optimum During
Normal Operations
The comments were unanimously positive:
This is the No. 1 (out of 6) ranked incentive expected, e.g. operating at the optimum steam
temperature, excess air, etc.
This is probably the No. 1 (out of 6) ranked incentive expected. Also, often the constraints
will change in the direction of the optimum once historical data is collected. This has
happened on boilers with excess air constraints.
Cost savings in the process industries have been very large, in terms of production increases
and/or cost savings at constant production. For instance, refinery savings due to DCS
average about $.12/bbl (6% of total production costs).
Boiler jobs (that we have done on DCS) have achieved significant savings in energy, e.g.
excess oxygen prior to DCS averaged 6-8%; every DCS job achieves 2% or lower.
This has been significant for our boiler control excess air is now much less variable. We
have changed from O2 control to O2 with CO trim.
We have experienced on a utility boiler DCS upgrade: 1.) a heat rate improvement of 1.5%,
of which 25-50% is directly the result of the DCS itself, and 2.) a reduction of 200 kW in
auxiliary power, through the elimination of relays and other control equipment, and through
more efficient operation of ID/FD fans.
One interviewee did sound a note of caution --For about 75% of the DCS applications we
will indeed achieve decreased process variation. This will often be primarily due to the increased
visibility of the performance of the field elements of the loop and their subsequent repair or
replacement. For about 25% of the DCS applications we will often get worse loop performance.
This is often due to misusing the vast, available, library of DCS logic structures without
understanding the basic principals of control loop design. He added that this problem gets
corrected in the field, and with experience, becomes avoidable through the Controls
Requirements Documentation.
2-10
Experiences with digital control systems show significant energy and material
savings and/or effective capacity increases, resulting from operating a fossil unit (or a
closely equivalent process) with less variation, and closer to the optimum, during
normal operations.
This is the No. 2 (out of 6) ranked incentive expected. Although very few upsets are
preventable via the control system, a digital system can significantly lower those that are
preventable with a control system.
This is the No. 3 (out of 6) ranked incentive expected. The number of boiler trips has been
demonstrated to decrease significantly.
Abnormal Situation Management (ASM) group of oil/chemical industry projects that one
third of all upsets are preventable with a digital control system, if the required software is
included. Software includes ordinary control point alarms, plus programs that look for
historical failure patterns across several process variables.
The DCS has been of great benefit in post-upset diagnostics, particularly for analyzing the
causes of boiler trips and changing procedures. This has led to a 60-70% decrease in drum
level trips, identifying and using Best Practice and then configuring smart alarm limits,
based on the real likelihood that drum level is out of control. We have also identified the
fastest, safest boiler startups and shutdowns, and use Help screens to coach every operator
through the Best Practice.
This has been an important benefit of the DCS, particularly in minimizing the magnitude of
the effects of any upset. For instance, we use on-line help screens to give the operator
guidance on recovering from an upset.
This has been a significant benefit. Bark boilers routinely experience upsets in bark feed.
Using advanced control techniques, at one of our mills, has virtually eliminated drum level
trips.
With the DCS logic capability, the impact of load change upsets on a hard coal burning
utility boiler was reduced by a factor of 2. This translated into a benefit of speeding up load
change rate on the unit from 5 MW/min to 10 MW/min. Oil used to fire this boiler during
startup was also reduced by 30%, due to getting up to coal firing temperature 10-12 hours
sooner. Faster startups were attributed mainly to the operators having graphic displays of
motor interlock and startup sequencing information. Boiler trips have been reduced from an
average of 10 per year to 1 or 2.
2-11
These are impressive benefits. It should be pointed out that they were not achieved
by simply replacing all the analog loops with the equivalent digital algorithms.
Process models and transfer functions were developed. Model parameters were
estimated from data gathered on the system. Advanced control algorithms were
configured, tested, tuned, and re-tuned.
Manpower savings should not be a DCS incentive. Manpower shifts to higher level
functions should be expected. Retraining to accomplish this should be planned.
At a mill where we converted 5 boilers, there was one operator per boiler. After the
conversion, 2 operators ran all 5.
There is little doubt that the DCS has the capability to achieve high levels of
automation. Again, it will not happen by simply replacing the analog loops with the
equivalent digitally implemented loops and operator interfaces, and then using the
same operational procedures. Success here requires motivation for change from all
involved organizations. The system must be developed to allow the operator to
manage by exception.
This should be ranked last as an incentive. The requirement for operating during most of the
transition will mean very little space reduction actually realized.
2-12
Control rooms are becoming manpower centers, including maintenance and lab personnel.
Space will not go down.
No real benefits have been derived from the extra space. Typically left as rack space.
No benefit here.
We have put the created space, due to a DCS conversion, to good use at one mill. We used it
for control racks for new equipment, which, otherwise, would have required building a new
rack room.
The only tangible benefit identified was associated with the avoided cost of a new
rack room for a process expansion in the same area. The operator/technician
interactions resulting from creating a manpower center can be considered as an
intangible benefit.
This is the No. 4 (out of 6) ranked incentive expected. Repair parts cost more, but faster
repair makes a digital system more cost effective. For instance, when sootblower systems
were controlled by relays, heat rate penalties (during the typically longer repair times after a
relay failure) were significantly higher than with a PLC.
Maintenance savings can be expected to be 50% and greater after a DCS conversion (if the
conversion includes fieldbus).
No maintenance savings are produced; however, DCS operator action log makes the
maintenance dept. a lot more efficient, due to faster and more specific identification of
problems. Also there is a lot less hiding of operation errors under the category of
mysterious problems with the system. More maintenance training, PCs for maintenance,
Internet access, etc. are now required, costing more, but the increased efficiency more than
makes up for it.
We have not experienced savings here. The cost of a single repair of a DCS component is
actually higher than with panel-mounted instrumentation, but you do fewer repairs.
2-13
Maintenance is significantly improved with the DCS due to the increased visibility of loop
performance. Weve had process control engineers offsite observing loop performance,
identifying valve problems, and initiating maintenance. This was not possible without the
DCS.
For a station with 2 units, there was a savings of $100,000 per year on recorder paper, due to
installing the DCS. This was typical of many small cost reductions.
Digital control systems do not lower the direct cost of maintenance of the logic
elements of control loops, they typically increase it. However, they provide the
opportunity for dramatically improving the effectiveness of total control loop
maintenance, which shows up on the bottom line as an improvement in overall
plant efficiency.
This is the No. 3 (out of 6, 1 highest) ranked incentive expected. One of the biggest benefits
gained from a DCS conversion is the collection of historical data used to determine the root
cause of problems.
DCS can produce OSHA compliant logs on everything that happened. The historical data
can be used to diagnose problems and get back on line faster. For instance, in one case a
reactor exploded. All process conditions and operator actions were reviewed from the DCS
historical data. The analysis showed a metallurgy failure as the cause.
The documentation (DCS logs) produced after a safety or environmental event are invaluable
in dealing with organizations outside the mill. Having coordinated, time-stamped event
logs, showing exactly what happened, helps meet several objectives very quickly.
The ability to go back and diagnose problems from the DCS log has been helpful.
One of the biggest benefits we see is the use of advanced control techniques to get the most
out of boilers while staying within environmental (Opacity, NOx) constraints.
A digital control system is capable of storing, retrieving, and reporting process event
data with inherently higher reliability and validity than a manual log system. The
value of the benefits achieved will be highly site and situation specific.
2-14
This advanced control technique gives the most bang for the buck because it is included
with the DCS and is relatively easy to set up.
Seven-element feed water (drum level) control is becoming the standard because
configuration and tuning are easier with a digital system.
Feed forward is routinely implemented. We typically use 5-element drum level control.
We are doing this more with the DCS then we did in the past; but, it is not a substantial
difference.
One of the real advantages of a DCS. This will definitely enable minimizing process
variability.
This should be a real advantage for boiler operation in a dynamic process environment. Not
as important in base loaded installations.
2-15
2.4.2 Loop decoupling using process mathematical model, e.g. Dynamic Matrix
Control or Inferential Model Control
In the right applications this can be beneficial. Analyze problem requirements first before
selecting a tool to use.
Multivariable control, using mathematical models for decoupling, has become the state-ofthe-art with digital systems.
This is less important than No. 1 (Simple PID with Feedforward) above.
On our fuel/air controls on combination boilers using bark, we need the heating value of bark
in the calculation. We use a mathematical model to calculate this based on several process
variables.
The use of model reference control algorithms has increased significantly with the DCS. We
use the historical data acquired through the DCS to update the models. A typical model is
reaction kinetics for air oxidation.
I am not aware of DMC in Power, but it is widely used in the Refinery Business. A
refinery I was involved with (my role was small), stated that the whole capital cost of the
DCS & Advanced Control was paid for by the advanced control (multi-million). The main
role for the DCS in this case was to give reliable repeatable control.
This is used. Also, mathematical process models are commonly used for data reconciliation,
i.e. determining when a process variable is in error and determining the correct value from
the remaining measurements.
Were getting a big benefit from calculating drum level alarm and trip points dynamically,
because shrink/swell effects vary so much with production level.
We use mass and heat balance models to determine upsets and then compensate in our
controllers.
Having the ability to change loop performance during upsets versus normal load variation
is a significant advantage in operations with dynamic operating modes.
2-16
This will definitely be beneficial to power plants. It is included with DCS and is easy to
setup.
This will have the biggest impact on the typical power plant.
This is used by power plants with a more conservative approach; those not willing to use
multivariable control.
Several control loops are set up with adaptive tuning constants. Drum level controls and pH
control loops are examples.
Being able to handle non-linear process dynamics through the operating range is a real
value for plants with variations in operating modes.
There has been a lot of success with this in predictive modeling, e.g. CEMS prediction or
NOX prediction.
This technique has been proven to be robust and reliable in the process industries. For a
power plant a typical application would be the NOX/Heat rate/Production optimization
control. (For a given NOX and Production constraint, control steam generation to the
lowest heat rate). Also, scrubbers and ammonia injectors both have non linear transfer
functions, and they are tough to model using first principles. Neural nets handle these
problems with ease.
This technique has been used for analytical variables that are difficult and/or expensive to
measure, e.g. the color of fuel oil is controlled by its correlation with temperature, pressure,
and flowrate, rather than with a possibly problematic color spectra-photometer.
We are using this for emissions monitoring, implemented as a DCS add-on package.
True NN control is very difficult to achieve. Not sure if would be worth the efforts for
powerhouse operations. Dynamics are pretty well understood.
There have been some limited successes with this technique. A superheater temperature
control loop preventing overshoot is an example.
2-17
Fuzzy controllers, in general, offer little improvement over traditional PID's, and none over
model based controllers like the Smith Predictor. However, they do seem to do well when
there is a large deviation from setpoint (like when a large setpoint change is made) AND
when the process has long time delays and dead times. This is due to the inherent non linear
control action of the fuzzy controller. Again, a PID can be made to behave this way by
adaptive tuning or even simply squaring the input error signal.
Not typically used in final control element loop, but can be used in problem diagnosis.
Weve tried this, but have not gotten anything productive out of it.
We are using this technique, combined with predictive models, extensively for optimization
in our energy management supervisory systems. One mill is using it to continuously
optimize the make/buy power decision. The system is credited with savings of $50,000 per
month. Another mill is using it as the plant master over 4 power boilers feeding the same
header.
Fuzzy logic would be of definite benefit. Vary performance expectations based on operating
conditions.
This is an early warning technique for identifying process upsets, prior to individual point
alarms and should be seriously considered.
We use this for on-line variability analysis of process variables. This identifies loops
performing abnormally.
This is being used by others for control loop performance measurement, but we are not using
it.
Not as relevant for specific process parameters but worthwhile for overall key measures like
efficiency, daily operating rates, downtime tracking and analysis.
Advanced control algorithms are definitely a key factor in achieving the benefits
cited for a digital control system. There are several choices available that overlap on
the objectives that they will meet. There are advanced algorithms that are built into
the typical DCS, and there are add on packages (software, hardware, or both)
available for others. The interviewees all concurred that it is best to first define the
control objective, and second evaluate, select, and test the advanced tool to achieve it.
2-18
If the definition includes low-level sequence controls, then sootblowing sequencing is the
probably the most beneficial on digital systems, due to the capability of optimizing the steam
usage to the determinable heat transfer coefficient. Most classical batch controls are not
beneficial to power plants because of the infrequency of equipment startups and shutdowns.
It is not worthwhile to develop a batch program unless the procedure is performed at least on
a weekly frequency.
Benefits are real in terms of attaining fast, reliable, startups. Startups of pulverizers, boiler
water feed pumps, and turbines are candidates.
One button equipment startups in the oil/petrochemical plant are commonplace. A major
refinery will bring a distillation column on line completely under batch control.
This is used on power plant control systems in the middle east, but not typically in the U.S.
I would recommend batch controls for: Ash removal, conveyor systems, boiler startups and
shutdowns. Even if operator manually executes each step, putting up Best Practice
prompt messages will improve operations.
We have achieved some major benefits from batch controls; however, there is a common
problem that must be overcome defining the procedure in standardized program. We
recommend using IEC 1131-3 to define the logic.
Utility processes arent recipe driven. One button startups eventually impact operator
skills. More value in help screens for startups. Conveyors and fuel handling processes could
be opportunities
This has had a significant impact for chemical plants, particularly on troubleshooting from
historical data. At a major chemical company, a certain reactor startup requires 8 hours.
The cost of any problem that trips out a reactor is very high. Trips are analyzed and
measures are taken to avoid future trips from the same problem class. The use of
mathematical programming on computers to minimize production costs is also of significant
impact.
Processing plant raw material inventories have been reduced significantly through the use of
business information processing systems tied to DCSs. Maintenance on valves can be
scheduled based on accumulated valve travel distance, rather than time. This is more
correlated with packing wear and saves significant money. For fossil units, the actual,
realtime, heat rates of the units, rather than assumed heat rates, should lead to better
dispatch.
Major impacts have been 1.) Increased availability of plant process data throughout the
organization, 2.) Increased reliability of numbers in the historical data bases, and 3.)
Increased speed and frequency of reporting allows more timely actions. On line production
scheduling in oil refineries has led to about a 4% improvement in operational efficiency
Availability of realtime plant process data in the system for various economic analyses lead to
1-2 % reductions in operating costs.
We have not integrated our DCS with a separate information processing system. Boiler
house management is implemented on the DCS vendor package. Historicizing of data is
within the DCS.
All of our mills have either (OSI Software) PI or (AspenTech) CIM/21 and it is used
extensively. Process control engineers at Corporate Engineering routinely log on to these
systems to monitor control performance.
Managers are used to MS applications such as Word, Excel etc. They demand that plant
data is available to them in this format. Trip analysis and other plant information is also
important.
2-20
Real time cost data to the operators. True dynamic costs to the decision makers. Accurate
performance tracking. Uptime analysis and troubleshooting. The potential for system wide
optimization in real time.
The next subject on this topic dealt with the extent that realtime plant data is used by
plant and general office management. Comments:
Terminal access of process data by a chemical plant manager is rare. Definitely used by plant
process engineers, maintenance engineers and supervisors, and plant business managers.
In the oil/petrochemical industry, the use goes all the way to the top of the company. The
president of a major oil company has a PC on his credenza and typically monitors total
amount of production instantaneously online.
Plant managers may use the system to monitor over all performance. Direct observation of
process data by corporate officers above this level is extremely rare.
Mill managers are definitely monitoring key process variables around the mill. V.P.
responsible for corporate energy use is constantly comparing boiler operations across similar
boilers in the company. Poor performance gets attention very quickly.
Higher level process people responsible for each production unit access process data in
screens and reports implemented on the DCS.
Operations management typically monitor trend graphics on key variables in their areas.
Unit managers will generate reports on the previous days process data logs. The mill
manager will typically monitor trends on key variables for the entire mill.
Real time process data is used by production engineers to help optimize processes. Mill
management looks at key operating info to track plant performance. E.g. Steaming rates or
minimum cost fuel flows. General office folks have yet to understand the value.
Finally, under the topic of integration of process control systems with IP, the
interviewees were asked about experience with a process industry equivalent of power
utility real time dispatch (i.e. setting of plant output directly from a remote central
office)
Its being done. A major supplier of nitrogen controls nitrogen/air liquefaction and supply
for hundreds of over-the-fence plants from a single control room in New York State. The
Plant Master controller in a multiple boiler facility is also analogous.
Production rate from off shore oil wells is typically dispatched from a central office.
Several utilities use real time dispatch for fossil plants, e.g. Baltimore.
2-21
Paper mills have fairly frequent upsets in steam demand, e.g. due to a break on a paper
machine. Depending on the power plant configuration, a number of loops will be in
supervisory control mode to a plant master which will make adjustments to set points
immediately when a break occurs to minimize venting steam.
We are using energy management supervisory systems external to the regulatory control
systems in the control rooms. One mill is using it to continuously optimize the make/buy
power decision. The system is credited with savings of $50,000 per month. Another mill is
using it as the plant master over 4 power boilers feeding the same header.
The paper mill plant master examples are realtime and remote supervisory; however,
the remote aspect does not include common carrier telephone connections.
The integration of digital process control systems with information processing
systems is just about universal in the process industries. Some plants try to stretch
the DCS to be both a process control system and an information processing system,
but this is definitely not the trend. Making the process variables visible beyond the
control system in realtime has definite value in almost all business decision making
functions. The example given of the use of external energy management systems to
optimize power plant setpoints is also becoming the norm.
This doesnt happen, companies cant afford the time that it takes.
Most companies will do an audit after the first two or three major DCS conversion projects.
The results usually are that the savings are much bigger than used to justify the project.
Then they dont bother with audits any more.
My company audits every project, and every DCS conversion project has met objectives and
produced the returns necessary to justify. Justifying a DCS conversion project today on a
boiler is really grabbing the low-hanging fruit. The only negatives have been unfair
criticism of inefficient operation of the boiler, prior to the DCS, e.g. why were you operating
so inefficiently for so long?
2-22
We have now converted all of our 8 control rooms to DCS. After the first few we did audits.
Then the need for audits disappeared.
When we converted a Kamyr (continuous pulp mill) digester to DCS, it was justified on pulp
yield variation improvements and on production increases. This was confirmed with an
audit.
Post installation audits of benefits are not routine. Where they have been performed,
realized benefits have met or exceeded the plan.
2-23
3
DIGITAL CONTROL SYSTEM ARCHITECTURE
The primary issues in the following three topical areas on digital control system
architecture were discussed with the interviewees:
1. Pure DCS vs. DCS/PLC Hybrid
2. Operator Interfaces
3. Use of Smart Field Elements and Field Device Networks
Decision driven primarily by cost, i.e. PLC I/O is cheaper than DCS I/O; however, cost of
system integration hardware and software will negate some or all of the savings.
Plant labor organization may require separation of discrete devices from instrumentation
loops. Although PLC may appear less expensive, other factors may over-ride, e.g. DCS has
better diagnostics, PLCs are not redundant, and two sets of tools and training are required
for the hybrid.
This will soon not be an issue. New systems will be true hybrids with instrumentation
controllers and discrete logic controllers on the same networks.
Cost often drives the decision on discrete logic to a PLC; however, the hidden cost of the
DCS/PLC gateway (hardware and software) will negate the savings.
3-1
The weak point of the DCS/PLC hybrid is the interface initial cost, programming cost,
operation -- How do you update the software on one side of the interface without updating
the other side? The strong point of the hybrid is the range of PLC solutions available.
Major factors are: Maintenance troubleshooting capability, best fit with skills/experience of
the engineers who will be implementing the system, best fit with process control
requirements of equipment , e.g. safety system may require a level of redundancy not
required on other DCS loops, and the discrete I/O capability and cost on the DCS.
Electrical technicians typically do not want to troubleshoot from boolean logic; they prefer
ladder diagrams. Some applications require the high speed of a PLC, e.g. counting sheets of
pulp coming from a cutter/layboy. Specialized systems, e.g. safety systems, burner
management, are often much more efficient on a PLC.
Some of the issues are: User preference, availability of redundancy, cost, maintenance logic
documentation requirements, configuration skills available, loop isolation requirements, and
availability of interface to third party equipment.
For process systems we want digital and analog information as closely coupled as possible,
including safety, interlock, and process information to the operators. Therefore, a DCS, with
discrete capability integral, is recommended. For processes that are discrete only and
interrelated information is not needed, PLCs are recommended. Common configuration
skills and common maintenance technicians for a pure DCS system is a plus.
When asked what the decisions had been for the installations they had been involved
in, the comments were consistent they had indeed gone both ways, depending on the
specific installation.
The interviewees were then asked for their advice for a specific type of installation, a
fossil utility plant.
Consider three factors: 1.) Degree of integration, 2.) Speed of discrete device processing
required, and 3.) Relative number of discrete to analog I/O points. With a low degree of
integration, high speed required for discrete logic, and a relatively large ratio of discrete to
analog, then put in PLC. Otherwise, put discretes in the DCS.
Consider control networks, such as Foundation Fieldbus and Control Net, with flexibility to
handle any mix of I/O types and logic functions.
Try to use DCS only, but dont make it a bigger issue than it is.
3-2
Consider two factors: 1.) Capability and job definition (per union contract) of maintenance
and engineers, and 2.) the level of information about the discrete I/O necessary to display to
the operators.
For the typical discrete device requirements on a boiler, we would recommend using the DCS
to do all logic except the burner management system.
For a fossil utility plant, unless there is a labor contract issue between
instrumentation and electrical maintenance, there is general concurrence to use the
DCS-only architecture.
Customers want PC based hardware platforms, but are typically not concerned about the
underlying operating system.
Vendor capability with the software is the primary factor. Select vendor first and then use
his best console software option for the job.
The overwhelming preference for NT is driven by the desire for interconnectability and the
cost/lack of satisfaction with proprietary operator interface solutions. The only negative is
the occasional screen lockup problem.
3-3
Cost is a big factor. DCS proprietary control stations were costing $50,000-100,000 each
(list). The use of commercial, PC-based hardware has lowered this to less than $15,000.
Operators also like using 21 CRT screens.
Most of the decisions have been vendor driven. Lots of attempts to try alternative
approaches like Wonderware. Again, stick with one vendor. Drive them to cheaper, more
screens, easy configuration, transferable configurations between generations of op stations,
better alarm management, a structured revision change process, screen redundancy, etc.
When asked what the recent decisions had been for the installations they had been
involved in, the comments were somewhat varied, primarily due to the represented
vendor organizations in the survey:
NT
Demand for touchscreen is declining. Demand for console furniture is declining. Customers
want the lower cost and flexibility of ordinary PCs on a desktop with a simple pointing
device.
The decision for our company is NT. Microsoft has the motivation to fix the lockup problem
with Release 5, and we expect that they will.
Considering the factors, we typically stay with the vendors standard console models, in his
proprietary design.
The interviewees were then asked for their advice for a decision made now for a fossil
utility plant:
Select NT, but use enough operator stations to allow for software failures.
Define the functional requirements of the console system before selecting a network
architecture: 1.) Required alarm handling rates are critical, a boiler feed water pump trip can
put 500 tags into alarm. 2.) Screen call up time and refresh time is important during an
abnormal event. 3.) Redundancy, in terms of the number of screens affected by a single
component failure, is also a factor.
3-4
Go with NT.
Use NT.
Use commercial boxes, redundant screens to assure operator visibility, cheap per screen
costs, networked views. Use NT. Unix is on the way down. NT reliability at the operator
level is acceptable. More understanding by on site resources, common boxes and operating
systems.
Select NT. If the DCS vendor does not offer an NT solution, consider carefully the
need for future modifications to the system and the future availability of spare parts.
To increase flexibility, use standard desktop NT workstations.
There should be an evolutionary approach. The operators should have the major input. They
should be allowed to evolve the graphics into what works best for them. Typically they will
evolve to very complex screens with a large number of update points and navigation keys to a
lot of other screens.
Start early by having the operators review the vendors Wedding Book of screens from his
past fossil plant jobs. It is important to setup a controlled procedure that develops peer
review of operator screen suggestions and incorporates the best ones.
Operators should design the graphics with strong guidance from human factors specialists.
There is a lot of research on ergonomic factors in graphics design and navigation.
Find willing operator, remove from normal duties, send him/her to DCS school, and then
team with maintenance and DCS configuration engineer to develop all graphics. This
achieves the best buy-in by the operators.
Use a tremendous amount of operator input for layout, particularly from the operations
foremen. Our operators typically control from process graphics. We maintain plant
graphics standards for color coding, line styles, etc.
3-5
Systems integrator should make his recommendations based on previous jobs, then operators
should make suggestions and changes. We recommend the use of site standards for graphics.
If site has no current standard, start with the ISA standard.
I can say without qualification that the best graphics have been developed by operators.
The best system I have seen is that one or two operators are taken from shift and designated
as operator trainers. They are then responsible for all operator interfaces, including design,
implementation, maintenance and the creation of the standards. There are many standards
for graphics, but I have found that this is always a contentious issue, so site standards are
developed. The important thing is that the operators develop a sense of ownership of the
operator interface.
Spend some time here. This is a key for the future. Engineers dont do this well. Need to
give operators a better view of the whole process. This is the disadvantage of DCS over the
old control panel if its not done well. Put in one big screen in the control room. Get
innovative.
Each operator needs 4 CRTs. One operator can be assigned 500 control loops + 500 discrete
points.
Each operator needs 5 CRTs, 4 for process equipment control and 1 dedicated to the alarm
summary. One operator can control 2 fossil units, with 12,000 console tags per unit. This is
happening now at Arizona Public Service.
Each operator needs a maximum of 3 CRTs. One operator can be assigned 200-500 control
loops.
For fossil plants, operators typically do not rely on screen navigation and expect 11 screens
10 for process equipment control and 1 dedicated to the alarm summary. For equivalent
equipment in the pulp and paper industry the same operator would have twice the equipment
3-6
responsibility and half the CRT screens, but this would not be acceptable in the utility
industry.
You need a minimum of 2 screens per operator; however, for one boiler controlled by one
operator, you need 3 screens. One operator can easily control 400-500 loops + associated
discrete devices. We have control rooms where one operator is controlling 1200 loops.
Our boiler house system has 6 screens for 4 boilers. Typically 1 screen is dedicated to a
custom overview display and 1 screen is dedicated to area trends. We have one board
operator and 3 rovers. There are about 500-700 loops in the boiler house.
Each operator needs 3-4 screens. One operator can control 4 power boilers, 300-400 loops.
Typically for a single Unit, (say 300-600 MW), there would be three or four sets of operator
stations, with stacked CRTs, i.e. two CRTs on top of each other. One operator can only
handle two sets, the others are for assistance when there is a plant upset. Normally one
operator would handle one Unit, with a supervisor that can help in a plant upset. For
smaller Units, one operator may handle two Units. Usually each Unit would be on a
separate communication system so that there can be no inadvertent operation, or a failure in
the communication system will only affect one Unit.
The process area and process dynamics are the key, not a loop count. One operator can run
many processes when the plant is running well. Key issue is startup and upsets. Control
screens should not be a limiting factor. They are, or at least should be, cheap now. 4 screens
minimum.
Each control room operator requires 3-4 screens and can handle approximately 400
instrumentation loops plus associated discrete devices. If these guidelines are
significantly different from perceived requirements, then other factors should be
reviewed, such as the Help screens, navigation scheme, alarm management
configuration, level of automation, particularly for preventing and recovering from
upsets, and level of operator training.
For a typical installation today, 75% of transmitters are purchased as Smart. The benefits
of ease and speed of setup more than justify the $40 per instrument increase in cost.
All transmitters today are purchased as smart. Smart valve positioners are far less
prevalent.
Wherever possible smart transmitters are applied due to the benefits of: accuracy, can be
located in areas which are not easily accessible to instrument technicians, can perform
diagnostics and calibrated from a central location (either PC or DCS consoles).
We are typically putting in all smarts, transmitters and valves. Maintenance is driving
this decision, and they feel that it is justified.
For our typical installation, 80% of transmitters and 50% of valves are smart. Not all of
the transmitter types we use are available from our vendor as Smart.
Our company standard is all Rosemount smart transmitters. We do not use smart
valves yet. We use the Entech valve standard and want to make sure that the smart valves
available will also meet the Entech standard.
Use smart transmitters, except where communications time affects loop response.
Be careful of aliasing and process dynamics issues when selecting smart transmitters.
Smart valves on all key process loops. Very little cost difference with smart sensors anymore.
Easier to use one transmitter to spare multiple locations. Can do press./ temp compensation
at the transmitter.
Smart transmitters are prevalent. Smart valves are in the process of being accepted,
and they offer major benefits in communicating valve condition and position history.
3-8
Have applied Smart Transmitters on a Pulp & Paper application where the DCS and
transmitter manufacturer were the same. The transmitter data was available on the DCS and
could recalibrate from the DCS console. A separate asset management PC was not required.
Had the DCS not been capable of direct access to the transmitters, a separate PC with asset
management software would have been applied due to the benefits of smart transmitters.
We have implemented the Loveland (Honeywell) network at two mills. Both are pleased with
the benefits.
We have implemented asset management software. Its still in the developing potential
stage. However, to get maximum control performance and therefore optimum process
performance the valves and sensors must be performing well. A predictive maintenance
approach is needed.
Have not implemented Fieldbus technology nor do I have near term projects that I can apply
it to. However, I do believe Fieldbus technology will be implemented in the near term and I
will follow-up on its application.
We have only put in one Foundation Fieldbus system. It is on a lime kiln, 25 loops, and
installed about 3 months ago. It is performing well. We would consider Fieldbus for a boiler,
but it would be a very careful consideration.
Our corporation is presently studying fieldbus. We dont think they will commit to it at this
stage of its development.
This is coming, but we will not be using for the next 2 years, while the initial bugs are
worked out. We dont see any advantage if using existing field instruments and existing
field wiring.
We have not and may not soon. However, in thinking about a control retrofit, especially
from a pneumatic system, this should be closely considered.
3-9
Foundation Fieldbus is finally a reality and will likely replace all 4-20 ma wiring
between smart field elements and the DCS controllers, in the near future, for new
installations. Few DCS vendors have complete, field proven, fieldbus ready systems
right now. It is a little too early in the technology for application to fossil utility
boilers, but this could change quickly. Even if the technology is not a concern, a
retrofit of an analog instrumented boiler with fieldbus is probably not justifiable.
3-10
4
DIGITAL CONTROL SYSTEMS SPECIFICATIONS AND
VENDORS
The typical procedure for acquiring a digital control system is to develop a specification
and qualify vendors who can reliably provide the equipment and services that meet
that specification. After qualification, cost becomes a decision factor, but certainly not
the only factor.
The scope of DCS vendor responsibility can range from Hardware and System
Software Only through Full Turnkey Implementation. Other vendors, consultants,
and contractors may be involved.
The only comments listed in this section are by individuals presently affiliated with
companies that are not DCS vendors.
We would only use a specification if we were choosing between several DCS vendors. Once a
DCS vendor is selected for a site or company, the specification is a waste. The specification,
if used, should only state functional requirements. Where the DCS vendor has been selected,
we typically issue a document defining: 1.) the choices of DCS options, 2.) system block
diagram, 3.) I/O count, and 4.) Quantities of other equipment.
No. We have selected our DCS at each operating location. Once that is done, we work with
the supplier on all new projects to develop the required system architecture as well as any
migration planning for the installed base.
4-1
Our company qualifies DCS and PLC vendors at the corporate level on the basis of product
quality, price, and service. All sites either use one of the vendors qualified by corporate or
justifies their decision.
Each site makes a decision on the DCS vendor (or vendors) with the best combination of
price, delivery, and service, and sets up a purchase agreement with that vendor. There is a
corporate agreement on PLCs Allen-Bradley is the standard.
Almost all of the systems can do the job. Many of the decision points should be in what you
expect from your vendor after you buy. Engineering support, process optimization support,
maintenance, spares management and inventory, etc.
Qualification is similar to other vendors, with the added complication that the
customer is typically heavily dependent on the vendor for support throughout the
life cycle of the product. This makes honesty, trust, track record, and experience in
the specific industry extremely important components of quality and service.
Interviewees were asked if they standardize on a single vendor for each (DCS, PLC) per
site. Per all sites?
No. Corporation has qualified 3 DCS vendors and 2 PLC vendors. All models by a vendor
are considered qualified once a vendor is qualified.
Yes, Corporation has one preferred DCS vendor and one preferred PLC vendor for all
locations. Not using the preferred vendors requires a detailed explanation.
Typically one DCS vendor per site, but a few sites have 2.
We standardize at each site; however I would recommend that you standardize for all sites.
To much synergy lost if you dont. Common configurations, procedures, hardware,
engineers, etc.
4-2
Process expertise is the main factor. We also consider success with advanced control schemes
and staging capability.
Consider the resources (people and dollars) that you have available and the capabilities that
the DCS vendor and the A/E firm have to offer. We have used about all combinations, and
gotten good results with all. Best combination is for Owner to purchase DCS, A/E firm
install and wire, and Owner configure DCS in-house.
The skills that are needed and you dont have or dont want to keep on staff. Project needs
versus system maintenance and optimization needs. Turnkey projects versus EPC.
Proximity of your sites; shared inventories or vendor managed inventories. Competencies of
the vendors.
The next question asked under this topic was: Have you ever given a DCS vendor
complete, turnkey, full implementation responsibility for a retrofit to a panel-mounted
control room? What was your evaluation of their performance?
Yes, if the vendor is qualified. We did it with Foxboro one of our mills and the job went
extremely well.
We have given Bailey complete responsibility. The people assigned the job were not as
experienced as A/E engineers assigned to the same work.
Yes. Generalizations are never very good, but they dont have good project management
skill sets in general. Their expertise is in their systems. If you do it, pay close attention to the
specific people assigned to your project.
The last question asked under this topic was: What services do other vendors,
consultants, in-house employees, and contractors provide, and how are these services
defined, scheduled, and coordinated?
The following services are often performed by a specialized third party or consultant: Safety
audit, graphics building assistance, information system interfacing, startup/checkout
assistance, and advanced controls.
We use 3rd party specialist sometimes for advanced control logic development.
4-3
For major jobs we use the A/E for configuration, if the right people can be assigned;
otherwise, we will use a system integrator for configuration. For small jobs the mill will do
the configuration.
There are specific functions required for the scoping, design, installation, checkout, startup,
optimization, maintenance and management of any system. Who you assign responsibility
for those functions depends on many factors; cost, time, skills, focus, priorities, preferences,
career development, labor contracts, etc.. You must analyze each individually and within the
constraints of the current situation.
The key phrase in the above comments is if the right people can be assigned,.
Implementers of digital control systems in the process industries have learned how
to recognize qualified people. The vendors responsibility, or any other
organizations responsibility, will be dependent on depth of available, qualified,
people.
4-4
5
IMPLEMENTATION PROCEDURES
The success of a digital control system upgrade will depend on how completely,
accurately, and clearly the process I/O and logic requirements are defined. After the
digital control system is configured, the system must be verified.
The success of a digital control system upgrade will also depend on the training
provided to the operators and technicians. Training is also a key to operator and
maintenance buy-in and ownership of the system.
Loop sheets, motor elementaries, wiring lists, and some mechanism for documenting control
strategy, e.g. SAMAs or IEC-1131 documents.
I/O database list for DCS vendor to configure, logic diagrams, graphic sketches, and
connection diagrams.
P&IDs with interlock notes, loop sheets, wiring schematics, motor elementaries, SAMAs on
all complex loops, and logic narratives.
Loop sheets, motor elementaries. We prefer to use the DCS vendors equivalent of the SAMA
logic diagram.
Much would depend on the scope of the effort. Assuming a fairly major retrofit, a detailed
project scope. P&IDs. Instrument database. Instrument install specs. Valve specs.
Preferred vendors list. Panel and termination wiring diagrams. DCS architecture
5-1
Implementation Procedures
drawings. DCS equipment layouts. Control room and rack room drawings including
cutover plans. SAMA diagrams. Loop sheets. Loop narratives. Interlock diagrams.
Configuration standards if none existed. Configuration drawings. Operator screen layouts.
Failure mode logic.
There was one comment in disagreement with the others:
No specific documents are recommended. Survey the site with controls experts and let them
recommend the documentation requirements.
Users should verify any as is field documentation developed by the site survey team.
The connection diagrams would be field verified and the logic diagrams would be reviewed
with operations.
Review with operations and maintenance. Sometimes discrepancies are found during later
simulation, which is the same as a review.
Field check of existing wiring. Individual drawing checks by the designer. Yellow line
check of all drawings against each other. DCS I/O check. DCS configurations tests. Pre
start up inspection.
Implementation Procedures
Many DCS users require loop back type simulation to verify the DCS implementation of the
logic.
Simulation is required only if the logic is intricate or different. For ordinary loops it is not
required.
A factory acceptance test is performed at the DCS vendors facility. Every I/O is simulated.
The test is witnessed by plant operations to confirm that configured logic matches operating
conditions. This is the final review prior to start-up.
We do not use simulation at all to verify configuration. We check logic when the system is
wired to the real equipment and field instrumentation.
We prefer to verify with simulation within the DCS. For Honeywell, we typically have the
people doing the configuration write the simulation in CL [Honeywell Control Language].
Yes we use a simulator, but we also check each I/O point and the internal logic via external
test inputs.
There is no concurrence on the best method for verification of configured logic. Most
are using some level of simulation, but without specific testing procedures.
5.3 Staging
On the subject of staging, the interviewees were asked: Do you stage a DCS, either at
the vendors facility or at the site, prior to installing and connecting field wiring? With
one exception, the interviewees were in agreement:
Staging at factory.
Yes. Factory test to see that it powers up and works OK. On site tests to verify I/O,
interfaces, and configuration.
5-3
Implementation Procedures
Definitely do not stage a system. It is a waste of time and money for todays DCS
architectures.
Yes, check loop hardware and software integration in the field prior to startup.
Absolutely.
Yes, the use of a simulator for operating training is typical for a chemical process.
Yes.
Yes, we use a simulator in a separate computer linked to the DCS. The simulator is designed
to automatically adapt to any changes in DCS logic and graphics so that it will remain up to
date as the DCS is changed.
We have completed development of a training simulator for one of our processes, and we are
developing simulators for some others.
5-4
Implementation Procedures
Yes, we do this for about of all major projects. There is pressure to drop this because of the
cost.
Yes.
The next question under training was: Do you use self-directed training? With testing?
No, not unless you are trying to teach the process, rather than teach the control system to
operators who already know the process.
Yes.
Self directed training is not recommended, detailed training courses by an instructor are
recommended.
Yes. Operator solve upset scenarios on the simulator and are timed on how quickly they
solve the problem. Simulator has to be fairly close to the process to make this work same
graphics, same tags, approximately same process response.
It varies with the project and individual operations management interest. Mostly the
training is formal. Sometimes testing.
And finally on training: What is the most effective method of training operators?
Training technicians?
Process simulator on the DCS or interfaced to the DCS is the best way to train operators.
Training technicians is very difficult, but very important. This is often overlooked.
Process simulator on the DCS or interfaced to the DCS is typical. A further enhancement
being used in oil/petrochemical is to set up the simulator to track a real process and
automatically adapt its behavior to the real process.
Training on the actual operator consoles is the most effective rather than generic training
consoles in the factory. Factory training is more effective for maintenance technicians.
Operator buy-in is important. They participate in the design (P&ID reviews, graphics
development). They go to the vendor schools. And, they train on the DCS simulator.
For operators: we take them off the board, give them class room instruction, as well as time
for the self directed exercises on the simulator. For technicians: we typically send them off
5-5
Implementation Procedures
site to the vendors facility. There the equipment can be bugged and the student can
practice diagnostic skills.
We use the simulator problem solving scores to qualify operators at 2 mills, i.e., if they cant
pass the simulator test, they cant become an operator.
Train to use DCS first on simple loops and a few motors, so that the operator can adjust from
a panel to a CRT/keyboard environment. Then train on the complete process using the DCS.
Operators write the training manuals. Train on a simulator. Involved in PSI (pre startup
inspection). For technicians: Just train those folks who will be working directly on the
systems immediately after the training. Bring suppliers to the site and train on actual
equipment. Do it concurrently with configuration checkout, communication interface
checks, I/O testing and operator training.
Operator training on the same model DCS console and same graphics to be installed
is recommended. Simulation must be used to take the place of real boiler and real
I/O. There was no concurrence on the simulator requirements. Maintenance
technician training is best handled offsite at the DCS vendors training facility.
Yes, this is associated with poor alarm management, either within the DCS capability or
resulting from poorly conceived setup of alarm functions.
Yes, this definitely can happen. It is often due to poor screen navigation design and/or to
poor assignment of alarm groups/filters. Alarm priorities and filters need to be very carefully
considered.
Yes, there are 2 solutions to be considered. 1.) Artificial intelligence (expert systems) to
assist the operators, and 2.) Rethink the alarming scheme, particularly alarm cascading and
dynamic filtering. Our facility uses an alarming design criteria whereby not more than 3
alarms can cascade from a single abnormal event.
A considerable amount of time has to be allowed for training because each operator must be
comfortable with his ability to absorb data at his comprehension level and speed, the plant
cannot afford to have the operator be overwhelmed because of time constraints.
5-6
Implementation Procedures
This is due to poor system design, particularly graphics and alarm management. Need to
review both, if this is a problem.
This is due to creating too many alarms and setting the alarm limits too tight. Operators
will then inhibit or disable alarms, which creates a problem later. We require operations to
print out a list of inhibited alarms at the end of each shift.
Alarm overloads occur when you are not careful in setting alarm limits and creating alarm
states. Where retrofitting a control room with annuciator panel alarms, we will leave the
panel alarms installed and hardwire the supercritical alarms to the panel alarm.
Yes, the solution is careful design of the operator interface and the use of simulators for
training.
Yes. First, make sure they know their process. Then develop and implement an alarm
management plan. Install help screens to diagnose and solve problems. Present operators
with knowledge on the screen, not just data. Train well up front. Make navigation between
screens intuitive.
And finally: Have you seen any problems with mode confusion, where the operators
were not sure of which control mode they were in and how to quickly get back to a
familiar mode? If so, what was the solution?
Yes, this problem is best solved by color coding control stations so that the normal mode of
control is a specific color.
Yes, this problem is best solved by training the operators in console procedures.
Typically not a problem, but it would be solved by using color graphics to alert the operator
that the control mode was abnormal.
Sometimes operator problems are not exactly what the problem is at first defined to be. E.g.,
weve had color blind operators, operators who cant read, and operators who needed bifocals.
Our DCS has a normal mode switch on the keyboard that will put all loops in their normal
mode.
We use a mode control text graphic of loops, clearly differentiating the loops not in their
normal mode.
Several suggestions, including color coding of control mode and text listings of outof-normal mode stations, were offered for solving these two problems.
5-7
6
CONCLUSIONS
In the body of this report the investigators conclusions and recommendations were
shown in boldface, immediately following the interviewee comments for each topic.
This section of the report summarizes these conclusions and recommendations for the
four major topic headings.
Conclusions
beyond the control system in realtime has had definite value in almost all business
decision making functions.
The tasks required to upgrade to a DCS-based digital system, and the costs and risks
associated with each task were examined. There was nearly unanimous concurrence on
the order of march: Equipment must be purchased, the site must be prepared, and
control requirements must be documented. The new DCS must be configured with
logic verified, staged, wired, and checked out. The operators and maintenance
technicians must be trained. Although equipment purchase is a relatively high cost
item, it can be estimated accurately for an upgrade. Site preparation and conversion
coordination for an operating facility was identified as a critical task. Logic
requirements analysis, operator interface design, and modifications to the operators
basic approach to control were also identified as critical to achieving positive benefits.
6-2
Conclusions
configuration experience are a rare commodity, and are typically drawn from several
alternative sources to ensure job success.
6-3
A
SURVEY INTERVIEWEES
Name
Present
Company
Present Title
(time with
present
company)
Previous Titles,
Companies
Process control
engineer with
International
Paper (20 yrs)
Robert W.
Barber
Champion
International
Corporation
Senior
Controls
Engineer (8
yrs)
Murray A.
Champion
Yokogawa
Industrial
Automation
International
Power
Business
Developmen
t Manager
Jay D.
Colclazier
FisherRosemount
Systems, Inc.
Sr. Industry
Consultant (7
yrs)
Process
engineer with
Monsanto and
Celanese (14
yrs)
Don Frerichs
Elsag Bailey
Process
Automation
Director of
Applications
Research (35
yrs)
Chief
Applications
Engineer for
Pulp & Paper,
Iron&Steel
Dan D.
Glossner
Amoco
Chemicals,
Decatur
Plant
Maintenance
Supervisor
(20 yrs)
Process control
engineer,
project
engineer, same
plant
A-1
Survey Interviewees
Name
Present
Company
Present Title
(time with
present
company)
Previous Titles,
Companies
William J.
Harding
The
Foxboro
Company
Industry
Consultant
Power
Applications
(Recent hire
from Parsons
Engineering)
Assistant Chief
Instrumentation
and Controls
Engineer,
Parsons Power
Group (20 yrs)
Paul S.
Inglish
Honeywell
Industrial
Automation
and Control
Business
Development
Consultant
(20 yrs)
Process Control
Engineer to Sr.
Business
Analyst with
Exxon (8 yrs)
David J.
Latour
Union
Camp
Corporation
Senior
Project
Engineer (20
yrs)
Newtron Inc.,
Instrument
Technician
Chris E.
Rogers
Boise
Cascade
Corporation
Manager of
Electrical and
Process
Control ( 13
years)
ITT Rayonier
(10 years)
Fred Y.
Thomasson
Union
Camp
Corporation
Senior
Applications
Engineer (21
yrs)
Babcock &
Wilcox, Senior
Process Control
Engineer
A-2