Newsletter of the Computer Technology Division of ISA
MESSAGE FROM THE PAST DIRECTOR
COMPUTEC by Thom Walczak
The Problem With Progress:
I Computer Technology has grown by leaps and bounds. Since I purchased what I
consider to be my first commercially available PC in 1974, both the hardware and soft-
N ware capabilities of computers have grown exponentially. At the same time, more uses
have been developed for the PC than I ever could have imagined back then.
Intel “owns” the PC hardware platform today, after taking the lead away from the
original owner, IBM in the 1980’s. This doesn’t of course discount Cyrix, AMD, and even
T Microsoft from becoming larger future players. In any case, the speed and performance
of current and future products make my first industrial automation computer installa-
tions look pale. I thought I had reached the ultimate plateau of success when I net-
E worked a DEC PDP-11/23+ to 30 islands of automation in the early 1980’s. Today I am
considering even removing this personal achievement for me from my resume!
At least I should probably at least eliminate the portion indicating that I am profi-
cient as an RSX-11M+ user. While looking closer at the changes in operating systems, it
R was DOS that really launched me into the true PC computing environment in the
1980’s. I know a lot of my friends as well as many of you probably remember working
on a TRS-80, TI-99, Apple II, Commodore, or other computer before we paid attention
F to operating systems (I owned each of these types at one point). Writing my first word
processor and data base programs in embedded basic at least gave me justification for
having a computer. Actually, owning a typewriter would have been easier.
According to Microsoft personnel, DOS and Windows 3.x are dead. I just became
A proficient at using 3.x about a year ago! For that matter, I still know many people who
are running under DOS today. I am still proud of the fact that as of last month all of my
machines have now been upgraded to WIN ‘95. What!?, you say that Windows NT will
C replace my WIN ‘95! For someone who has invested both time and money into “Plug
and Pray” (pun) compatible hardware and thrown away all my packages of DOS and
Windows 3.x software and several I/O cards to take advantage of the WIN ‘95 compati-
bility, I am confused. What more capability could I use? For the industrial control envi-
E ronment, I certainly could always use control and networking functionality which was
easier to implement, faster data transfer, and more compatibility for file exchange. In
fact, my network server is required to be tied to the NT operating system capability.
I imagine that now that my headaches of trying to merge my new application pro-
In this issue... grams data into the WIN ‘95 environment may mean doing all this again in 1998 with
Message from Past Director.........1 “Cairo”? Or maybe the US Government will really declare Microsoft operating systems
Computec Goals & Needs ..............2 a monopoly? That wouldn’t hurt that much would it? We have enough incompatibility
Comic: “Stranger Than Fiction” ....2 issues within the Microsoft transitions today.
Measure of CPU Power....................3 Well it’s time for me to get back to connecting my new network up. It is a WIN ‘95
Best Paper ISA/95 Computec TCP/IP peer serverless design. I have a dozen or so systems, and I don’t want to tie up
Migrating Control Systems to a computer with running Windows NT just to allow me to have file share access capa-
Open System Platforms .......4-9 bility.
I’d like to hear from anyone else who has thoughts on the future of Computer
Call for Papers...................................10
Technology. I occasionally log on to Internet, and I have an account through
CompuServe. This has worked out well for me since I travel quite frequently and can
log on locally in various locations. You can reach me at firstname.lastname@example.org.
ISA/96 AND ICS/96 CHICAGO TECHNICAL CONFERENCE
Monday, Oct. 7, 1996 - Thursday, Oct. 10, 1996
COMPUTEC GOALS FOR 1997/1998
1. Restructure Newsletter 16. Establish Executive Committee, Director, Director-Elect,
A. Division-wide participation Program Coordinator, Review Coordinator,
B. Recruit a number of contributing editors Membership Chairman, Newsletter Editor, Web Page
2. Get Session Developers for Generic Sessions Editor
3. Seek Paper Reviewers 17. Develop a New Member Welcome Package
4. Conduct Division Survey to Determine Technologies 18. Develop Ties with 21 ISA Sections
for Paper and Tutorial Solicitation.
5. Construct a Web Page and Get a Web Editor
6. Maintain Membership Database by Name, Company,
We Need Your Immediate
and Specialty Assistance to Meet These Goals -
7. Establish a Database of On-Line Resources for
Computer Technologies; i.e., a list of Control System • Any on-line numbers/addresses to add to a
BBS database of computer technology resources.
8. Establish a List of Free Computer Publications • Information about free computer publications
9. Advertise for a Review Coordinator that you might be aware of.
10. Contact IEEE Computer Technology Division for • We need a REVIEW COORDINATOR.
Interest in Doing a Symposium • We’re seeking 4 Symposium Coordinators.
Possible Site: Portland, OR with Pentium Plant Tour
11. Get four (4) Symposium Coordinators Any contributions to the
12. Evaluate Level of Effort and Resources for Symposium
13. Select Direct-Elect as Early in Year as Possible
Newsletter are welcome!
14. Increase Organizational Membership Editors are needed in the following areas:
Write a Professional Society Benefits Events • Trends • Hot Tips & Shortcuts • Software
Letter and Send to Division and Fortune 500 CEOs
15. Assure Consistent and Dependable Services to Please contact Charlie Gifford
Members (Home) 301-753-6543 or (Work) 301-743-4771 x1895
Letter of Commitment for Newsletter Editors and or (E-mail) email@example.com
P Lee Mikles, JLS Software Services, 302-633-0914
T HE H ARD W ARE C ENTER
The Real Measure of CPU Power Dr. Steve Belovich, BV Technologies, Inc.
“How powerful is my computer?” This question is asked Table 1: “Livermore Loops” Normalized FP Operations
by nearly everyone at least once. Computer makers’ sales and
Real FP Operation Normalized FP Operations
marketing efforts have emphasized this issue above all others
in order to promote their individual machines. Usually, the Add, Compare, Multiply, Subtract 1
answer to this question is given in terms of MHz, MIPS, Divide, Square Root 4
MFLOPS or SPECmarks. Unfortunately, the real answer to this
question depends upon many factors. Further, these figures are Cosine, Exponent, Sine, etc. 8
often very misleading and may be poor guides for selecting a Like MIPS, MFLOPS depends upon the program mix and
computer. compiler optimizations and so is subject to the same errors and
What are MIPS? misinterpretations.
Most people know that MIPS stands for “Millions of What are SPECmarks?
Instructions Per Second.” What is less well known is that this The computer industry has recognized the limitations of
figure is a statistical average which is based upon test results MIPs and MFLOPS in evaluating CPU power. In 1989, the
from several programs. To measure the MIPS of a given System Performance Evaluation Cooperative (SPEC) released a
machine, a group of programs is selected which are deemed set of benchmark programs to measure CPU speed. These are
“representative” of the “typical program” that a “typical user” real programs which are intended to be representative of real
will run. The total number of machine instructions in these workloads and are selected to measure both floating-point and
programs (including loops. BIOS calls, etc.) is counted. The integer performance (SPECf and SPECint). The SPECmark fig-
programs are then run on the machine and the total elapsed ures are computed based upon the execution times of the indi-
time needed to execute them is measured. The total number of vidual programs. The SPEC benchmark suite has been widely
instructions is then divided by the elapsed time to get the MIPS accepted, but it still should be used with caution since it may
rating. not reflect the actual workload.
Clearly, the MIPS rating depends heavily on the particular
programs selected, the compiler and even the operating sys- Things to Look Out For
tem. Selecting a different program mix will also change the Peak performance has little to do with actual performance.
MIPS rating, often by a factor of two or more. The so-called “Perfect Benchmarks” (designed by the “Perfect
Using different optimizations during compilation (or dif- Club,” a consortium of universities and companies interested
ferent computers) can drastically change the distribution of in parallel computation) were run on a CRAY X-MP/416, an
machine instructions, particularly with regards to the emula- IBM 3909-600S and a NEC SX/2 by Lubeck, Moore and Mendez
tion of floating-point operations. This effect can even cause the . In each case, the actual MFLOP performance of the
MIPS rating to vary inversely with CPU performance! Perfect Benchmarks (summarized via the harmonic mean) was
When the VAX 11/780 was introduced by DEC in 1977, barely 1% of the manufacturer’s claimed peak performance.
benchmark tests indicated that it was about as fast as an IBM Another trap for the unwary is confusing clock speed with
370/158. Since IBM's marketing referred to the 370/158 as a “1- performance. Between CPUs from the same family, higher
MIPS” computer, DEC’s marketing called the 11/780 a 1-MIPS clock speeds generally mean faster machines. However,
machine. The 11/780 was so popular that nearly all manufac- incresing the clock speed from 33 MHz to 100 MHz will not
turers gave MIPS ratings for their machines that were relative triple the overall performance. The number of clock cycles per
to a VAX 11/780. Thus, if a machine was 10 times faster than a instruction for portions of the instruction set usually must be
VAX 11/780, it was called to 10-MIPS computer. Reality set in increased in order to achieve higher clock speeds. More clock
during the early 1980s when Joe Emer for DEC actually mea- cycles are thus required to execute a given instruction which
sured VAX 11/780 performance and discovered that it was bare- can negate a significant portion fo the clock speed improve-
ly 0.51 MIPS. Since then DEC has used the VUP (VAX Unit of ment. This effect can be seen in the PC market where program
Performance) to indicate speed relative to a VAX 11/780. execution times have not decreased in proportion to clock
MFLOPS stands for “Millions of Floating-point Operations speed increases.
Per Second.” It is measured in a similar way as MIPS, i.e., using So How Do I Measure Performance?
a “typical” program mix. Since the set of floating-point (FP) The short answer is that you can’t - there is no single metric
instructions varies across computers, (e.g., a Cray-2 has no which is universally appicable in all situations. A good rule-of-
divide instruction whereas the Motorola 68882 has divide, thumb is to interpret the SPECmarks (or other real-program
square root, cosine and sine instructions). “normalized” benchmarks) in terms of your specific workload. If your main
MFLOPS are often used. The the table below shows the nor- application is compilation, then emphasize compiler perfor-
malized FP operations used by the authors of the “Livermore mance measurements. If your application is numeric analysis,
Loops” benchmark programs. then emphasize floating-point and vector performance. In all
cases, use common sense and don’t be fooled by the marketing
BEST PAPER - ISA/95 NEW ORLEANS - COMPUTEC DIVISION
Migrating Control Systems to Open System Platforms
William Bruce Johnson, Marketing Product Manager
Fisher-Rosemount Systems, Inc., Austin, Texas
Abstract users and vendors of proprietary control systems relied on cus-
tom interfaces for sharing process information and integrating
Control system users are faced with the ongoing challenge
the various fUnctions of the company.
of keeping existing systems and themselves updated with
With emerging and defined industry standards in data
current technology in order to improve competitiveness and
communication, operating systems, and now fieldbus protocol,
profitability. New, open system platforms offer users the ability
the word proprietary can have largely negative connotations
to enhance control strategies, improve communications, share
when describing system technology. Standards such as TCP/IP ,
process information, and extend their usage of application
UNIY OLE, DDE, and ODBC are enabling users to “open” their
systems with common data access protocols and interoperabili-
While open systems provide these advantages over tradi-
ty among platforms and devices. And, fieldbus will “open” the
tional, proprietary distributed control system architectures, con-
field management level of control and provide both control and
strained resources and budgets force users to properly justify
non-control information to functions throughout the company.
and prioritize upgrade projects based upon derived profits. This
Open systems are characterized by their embedded tech-
paper examines some system migration considerations, and
nology standards, their ability to integrate easily with other sys-
presents an effective method for planning migrations of existing
tems and devices, and their ability to provide/access informa-
tion throughout an organization. Custom interfaces and propri-
Closed vs. Open Systems etary operating systems and protocols will become less used
It helps to provide some comparison between closed and and less needed in the fUture of open system technology. At the
open systems when trying to evaluate their competitive useful- business management levels, enterprise-wide client-server
ness. The evolution of control systems (Pig. 1) over the past sev- architectures are now being implemented which allow infor-
eral years gives a good comparison between open and closed mation to flow freely to and from open control systems.
system technology. User Requirements in the Process Manufacturing
Users’ requirements and companies’ business objectives,
with implementing control system projects today, remain cen-
Reducing unit costs - pressure from competition, pressure
from customers, and pressure from rising internal operating
costs (e.g. health care, insurance, and regulation) continue to
push process manufacturers to explore opportunities for reduc-
ing unit costs.
Decreasing time to market - the ability to supply customers
quickly during smaller windows of market opportunity is
important to capitalizing on the profit phase of a given prod-
uct’s life cycle and to winning customer commitment. Shrinking
product development and manufacturing time from weeks to
Fig. 1 days can have a meaningful impact on financial performance.
Improving productivity, up-time and asset management -
from windows-based operator consoles to fieldbus instrumen-
As they were introduced in the early 80’s, distributed con-
tation, “open” system technologies are being evaluated more
trol systems (DCS) were comprised largely of proprietary-
closely for extending the use and maximizing the performance
designed controllers, consoles, I/O, and communication
of current and planned control systems.
devices, with one of the few standards being a 4-20mA signal for
Meeting regulatory compliance - with up to 20% of their
field devices. In designing their control systems, vendors were
capital being spent on regulatory compliance, process manufac-
not influenced as much by integration and information sharing
turers, as well as some control system suppliers, are at work
capabilities as they were by pure performance and computer
with various open system technologies to streamline compli-
technology advancements. Users, too, based their control sys-
ance procedures and costs.
tem purchases on these performance considerations, as well as,
Improving product quality - Quality is a competitive tool
the control application expertise of the vendor. As the process
industry moved toward computer integrated manufacturing, continued on page 5
Migrating Control Systems continued from 4 1. Define and link business, process, and system objectives
2. Evaluate alternatives and develop migration plan
which will continue to differentiate good products/suppliers 3. Document the plan
from inferior products/suppliers.
Increasing manufacturing flexibility - the ability to recon- Fig. 2 below provides a flow diagram for this migration plan-
figure and adjust for changing market forecasts, production ning process:
recipes, and feedstocks will promote flexibility and responsive-
ness. Maximizing equipment and facility usage with multiple
product manufacturing capability will beneficially impact
Each of the user requirements and business objectives
above can be positively impacted by open system technology.
Planned investments in open system technology offer some
dividend, the measure of which will vary by the application
and the company. The ability of ”open” process and plant con-
trol systems to provide and access useful information through-
out the organization will become more consequential as users
work to achieve these common business objectives.
The following quote from Fortune magazine (June 1993)
provides perspective on the perceived importance of informa-
tion technology: Fig. 2
“More than any other agent of change, information
technology is transforming the way business works. It is help- Step 1. Define and Link Business, Process, and System
ing companies get leaner, smarter, closer to the customer. Objectives
Those who seize the opportunities inherent in this revolution The first step in migration planning provides the basis for
are capturing. important competitive advantages Those who further project evaluation. All companies, small or large, have
fag behind are forced to scramble breathlessly in a race to catch business objectives. Process manufacturing companies, particu-
up, or die.” larly, will also have process and system objectives. Sometime
these three types of objectives are readily defined within the
Open systems embrace information technology. It will be company, and at other times they may require definition, clari-
helpful to now look at a method for planning open system tech- fication, and possible revision.
nology investments. Business objectives will be those objectives which directly
A System Migration Planning Method affect a company’s profit and stability. (The user requirements
outlined earlier can serve as a basis for defining specific business
Each user and/or company has various methods and pro- objectives.) Process objectives will be those which result in mea-
cedures that are typically followed when identifying, evaluat- surable improvements to yield, cost, quality, capacity, and pro-
ing, justifying, and gaining approval for control system projects. ductivity. System objectives will be objectives related to increas-
While these established methods/procedures are typically effec- es in system functionality, performance, capabilities, capacities,
tive and readily accepted, they can be inflexible and time con- and information or data communication. Examples of these
suming, and can paralyze efforts to make improvements, par- three types of objectives are:
ticularly with expansion, upgrade, and migration projects.
Further, some project approval procedures allow for much sub- Business objectives:
jectivity, and many projects get justified based on “wants” of the Improve quality of>product X to customers by 5% by 1996
user instead of the projects’ ability to meet company objectives. Meet 1996 EPA audit requirement
The migration planning method presented in summary Add two new Products at Hope plant within two years
below can be used independently of any existing procedures, or
can be utilized in concert with existing procedures to provide
additional support to the project justification process. While Improve environmental emissions auditing procedures
some project planning methods start with a process or system Increase capacity within Unit A to support additional pro-
issue, this method identifies and links the company’s business, duction
process, and system objectives as a basis to quantify, evaluate, Enhance SQC data collection process
and prioritize system investments. System objectives:
Depending on the size of the company and the size of cur-
rent control system(s) being considered, migration planning Add PC client to system with SQC application package
can and should be expanded to be a team responsibility as may Add 2 controllers and T10 to Unit A system
be required. As with any team activity, success will be enhanced Install monitoring package with automated rate reports to
by the selection of a management sponsor, who may in turn Console A
select a qualified team leader to lead the migration planning
When defining the business objectives, it is necessary to
work and schedule.
establish monetary values for each. Estimating the annual dollar
Migration planning consists of three logical steps:
continued on page 6
Migrating Control Systems continued from 5 The first task in Step 2 is to list the content of our existing
system products, along with the equipment options, revision
(or other currency) value of the objectives can be accomplished level, and any capacity/performance limitation.
by listing both tangible and intangible benefits. Tangible benefits Next, identify and document the technology solutions
can most often be expressed in terms of cost factor (savings/prof- available to accomplish your process and system objectives.
its) identified for the objective. Intangible benefits may require These solutions will include commercially available hardware
some thoughtful consideration, but can often be assigned a dol- and software products, as well as, application and process tech-
lar value based upon cost savings or cost avoidance. (Ex. - nology developments. Information about solutions may be
improved system reliability can reduce the losses associated obtained from control system vendors or other industry
with unscheduled shutdowns and can be given a percentage sources, such as, trade/technical publications and conferences.
allocation based on real or projected system failure data). Identifying and documenting solutions can be accomplished
These objectives are established in a hierarchical relation- easiest by considering each of your systems’ components
ship, with business objectives supported by the process individually, including:
objectives, and process objectives supported by the system Consoles Controllers Application Software
objectives. I/O Configuration Communication Devices
Next, linking the objectives will establish contingency rela- Maintenance Training Services
tionships between the objectives, and is essential in order to
substantiate their requirement and ensure their support. Benefits of these technology solutions will need to be deter-
Linking the objectives should identify and distinguish between mined and quantified as necessary in order that they can be
both “musts” and “wants”. A “must” is considered an absolute linked to support the objectives outlined in Step 1. The actual or
requirement in order to achieve a given objective, and a “want” estimated cost associated with each solution must also be
is indicated as a benefit/feature that is desirable, but may not identified.
necessarily be a “must” for realizing the given objective. After documenting solution alternatives, it is necessary to
The use of simple, two-dimensional matrixes (Fig. 3) or an link the possible technology solutions with the system objec-
electronic spreadsheet will be helpful to accomplish the linking tives to ensure their correlation and to begin identifying key
tasks. At this step, one matrix could be used to link the business migration opportunities and priorities. Here again, a two-
objectives with the process objectives, and a second matrix dimensional matrix (Fig. 4) will prove useful when linking
could be used to link the process objectives with the system system objectives and technology solution alternatives.
objectives. In this final linking process, which serves as an evaluation
of alternatives, migration opportunities and priorities will be
easily identified. In linking the solutions and objectives, oppor-
tunities will reveal themselves with their high return on invest-
ment potential (based on the values assigned in Steps 1 and 2
above), as well as, by their close correlation to company objec-
tives. Priorities can be established based on ROI calculations,
and by schedules and timelines associated with company’s
Step 2. Evaluate Alternatives and Develop Migration Plan
The second step documents your existing control system,
looks at the technology solutions available to accomplish the
business, process, and system objectives, and then develops a Fig. 4
prioritized plan for its implementation.
If your control system vendors are well apprised of our sys-
tem and process requirements and are looked upon as reliable
business partners, the vendors will prove to be a useful resource Step 3. Document the Plan
with this migration planning exercise. Otherwise, if you stay The plan itself should be written down, or documented, in
apprised of the technology solutions available for your control order to gain any necessary management approvals, communi-
system and can obtain approximate cost information regarding cate implementation details, and help keep the plan evergreen.
possible solutions, you will not necessarily require the assis- The work done earlier to establish links between business,
tance of an outside supplier.
continued on page 7
Migrating Control Systems continued from 6
process, and system objectives will serve as the basis for plan
justification. There is probably no better justification for a
system upgrade or migration than a direct link of the required
capital expenditure to the achievement of specific business
With migration planning, a schedule is also recommended
to keep the planning process focused and moving toward com-
pletion. Typically, if any planning process is conducted in
absence of a schedule, it is usually protracted, delayed, or
derailed in favor of other pending work. Regardless of the size
of the company of system being considered and to accurately
evaluate current technology solutions, no migration planning
process should extend beyond a six-month period. Fig. 5
While this paper presents the fundamental steps of a sys-
tem migration planning process, additional project justification To further evaluate the financial impact of these two alter-
work may be required by some organizations and can be read- natives, the user makes the following assumptions:
ily inserted into the process outlined here. 1. The migration option is forecasted to provide the user
Another important consideration with evaluating system with a 5 year competitive life cycle, while the replace-
migration projects involves the ability of a current control sys- ment option is forecasted to provide a 10 year competi-
tem to work effectively with newer, “open” system components tive life cycle. (The difference resulting from the age of
and platforms. A brief look at two system design factors, inter- the existing system).
operability and migratability, will be beneficial to any system 2. Both system alternatives, through the achievement of
migration planning exercise. business and process objectives, are expected to provide
System Migration Considerations annual net income of $200, 000.
Financial analyses of control system project have revealed Over its five year life cycle, the migration option yields
that evolving, or migrating, an existing distributed control sys- 44.4% return on investment (ROI%). Over its ten year life cycle,
tem to open system technology, when possible, can provide a the replacement option yields a 5.5% ROI.
better return on investment than executing a complete control The ability to evolve an existing system can make a
system replacement. significant difference in financial performance and project
Two significant control system design factors can help justification.
determine if evolving an existing system is possible and cost-
effective. One factor is called interoperability, which has been
defined] in the DCS world as the ability to control, view, manip- Open systems will mean more, better, and easier access to
ulate, configure, and/or diagnose an existing system from a new information throughout the organization. Advances in infor-
system or new system component. mation technology and open system standards, whether at the
The other design factor is known as migratability. field device level, the process management level, or the business
Migratability has been defined as the ability to convert existing management level, offer process manufacturers new opportu-
system equipment to the latest technology through hardware nities for improving performance. Further, information technol-
upgrades and software revisions. ogy is touted as the single largest agent of change in transform-
In essence, interoperability and migratability give the user ing the way businesses work.
the opportunity to extend the useful life (or life cycle) of an As with any capital authorization, migration project justifi-
existing control system. And by extending the useful life of an cation will typically include an acceptable return on the system
existing system, even with required incremental capital invest- investment and a direct relation to the company’s business
ments, users can often achieve a higher return on investment objectives. Migration planning should also include an assess-
than by a complete system replacement. ment and correlation of the company’s business, process, and
An example of the benefit of interoperability and migrata- system objectives, along with a linkage evaluation between the
bility can be found in the following project cost comparison system objectives and alternative technology solutions.
summary (Fig. 5). For this project, the user had specific business, Finally, the ability to evolve an existing control system to
process, and system objectives which necessitated substantial take advantage of open system technologies can provide sub-
functional enhancements. Initial project planning assumed that stantial benefits. The migratability of an existing system and the
a system replacement would be necessary, due mainly to the 7- interoperability between new and existing system components
year age of the existing control system. By evaluating the spe- can favorably impact the financial measurement of a migration
cific process and system objectives required to support the busi- project.
ness objectives, and by working with the vendor to identify
interoperable and migratable solutions, a migration decision
was reached and implemented at a savings of 75% over the
Call for Papers!
Society of Petroleum Engineers (SPE) / Industrial Computer Society (ICS)
June 8-11, 1997 • Dallas, Texas
We Are Seeking Papers/Tutorials Related to the Application of Computers in:
Data Analysis Modeling
For further information, please contact:
Charlie Gifford, Computec Div. Director
(Home) 301-753-6543 or (Work) 301-743-4771 x1895 or (E-mail) firstname.lastname@example.org
ISA DIVISION NEWSLETTER NON-PROFIT
DIVISION OF AMERICA
The International Society of
Measurement and Control
67 Alexander Drive
P.O. Box 12277
Research Triangle Park, NC 27709
ISA Online: http://www.isa.org