QuantifyQC HPMay92

MAY 1992
Hydrocarbon
Processing
Productivity
and Quality
Information system
A plantwide approach
was implemented
Plant automation
Develop a master plan
for greater success
Hidden benefits
Modeling can quantify
intangible benefits
Decision support
, Integrate real-time
plant operations data
Statistical methods
Use these to increase
on-line model accuracy
PINPOINT HYDROCARBON TYPES
UNDERSTAND ISO 9000'S APPLICATIONS
: TAME: TECHNOLOGY MERITS- -

PRODUCTIVITY & QUALITY
Quantify quality control's
intangible benefits
Combine the economic penalty model
for spec violation with conventional
benefit models for approaching the
spec, and SQC analysis reveals total
benefits can be twice as large as
quantified by conventional analysis
P. R. Latour, SETPOINT Inc., Houston
control systems, computers, software, optimizers, mod-
els, schedulers, blenders, surge tanks, and information
systems are made for improved process control (SPC/SQC).
In addition to equipment, major expenses accrue for engi-
neering, operations, maintenance and management for
process and quality control. Invariably these expenses
must be justified on improved plant performance. Improved
SPC/SQC performance is manifested in better yield of
valuable products, reduced energy consumption and oper-
ating costs, and sometimes increased plant capacity.
It is well recognized that justification ofimproved control
and information systems is a difficult task, but it is also
very important. The two major hurdles are assessing control
system performance (usually reductions in variability) and
the economic consequences of using the improved control
performance. The second issue is covered here.
Table 1. Typical examples for improved quality control
• Low sulfur fuel oil :0;1wt%S
• Gasoline octane <!94RON
• Heavy naphtha endpoint :o;390°F
• Fuel oil viscosity :0;640CS/122
• Crude heater temperature :O;730°F
• Distillation flooding <100%
• Compressor speed <6,200 rpm
• Gas oil 90% point, ASTM D86 :O;620°F
• Content of lube oil can <!1.0quart
'Bottom line.' What are the economic consequences of
using improved control? Many examples cite remarkable
payouts (less than one year) for im-
proved instrumentation and control
systems. Unfortunately, too many
examples cite poor investments. It is
essential to establish where the bene-
fits lie and where the biggest incen-
tives are. One can then determine
emphasis and priority for designing
comprehensive computer information
and control systems.
~ ~~~~~~~~~~_S~pe_c~
.~","",,-~ - ,,".,,=-.....,..~
~ , Reduce fluctuations S
Spec
Move average (setpoint) $' S
toward spec
Time-
Normal benefits procedure. Typical
criteria for product quality in a hydro-
carbon processing plant are given in
terms of a quality specification (spec)
limit. Common examples of product
specs are in Table 1. A product is
deemed acceptable if it is within the
T
he benefits of improved quality control can be dou-
ble what is normally quantified. The trick is to go
beyond typical methods, which consider only the
tangible profit gains from improvements in average yields
that result from operating closer to product specification.
Reducing quality variations alone is usually considered a
necessary step but produces no real financial benefit. But
modeling intangible losses that occur when products are
produced beyond specification limits provides a way to
define the intangible economic benefits from reducing
quality variations. Results in the oil refining sector show
that these benefits may equal those obtained from improv-
ing average product specification and yields.
Several questions stem from analyzing the benefits of
statistical quality control (SQC) and statistical process
control (SPC):
• Are there intangible benefits from
improved SQC, and if so, what is their
source and magnitude, and how are
they quantified?
• What is the penalty for violating
the specification or limit?
• On what basis is the specification
or limit set? Is it set properly?
• Why is it ''better" to operate on the
safe side and be conservative?
• Why are tangible benefits from
process yield, capacity and operating
costs easier to quantify than intangi-
ble benefits from customer satisfaction,
safety and maintenance?
Large investments for analyzers,
quality laboratories, instrumentation, Fig. 1. Control quality variations with time.

1.1 Mbpd
0.33 MMppdSpecification
Target limit
average -............
LGO
0.2%S
Specification
Target limit
average -............
'"c'0
Q.
<1>
c.
~1-~-----"-l----l7---="""1
'""0
Q;
.n
E
::J
Z
!'l
c:
'0 Base
Q. case
<1>
C.
E
~ ~~-------+------~~~~
"0
Q;
.o
E
::J
Z
Product quality, S
20 $b = 6.65 <!:/Ib
LSFQ
1.1%S
8.9 Mbpd~c ----'"
2.67 MMppd
l~M) = 5 ¢/Ib
Cost, e/lb = 5.0 e·n+ 6..65 (0.~3)=
5.185 ¢/Ib = Hi.55 $/b
Mbpd = Thousands of barrels per day
MMppd = Millions of pounds per day
Product quality, S
Fig. 4. Blending low sulfur fuel oil.Fig. 2. Quality distribution curves. Fig. 3. Quality distribution curves.
specification and unacceptable if it violates the specifica-
tion. These spec standards are typically written in the
legal contract for bulk supplies to represent the perfor-
mance of manufactured products.
The normal procedure for establishing benefits from
improved process control is based on steady-state con-
cepts.1-S The actual product quality will vary with time as
illustrated at the top of Fig. 1. This transient curve may
represent the output from an onstream quality analyzer on
the product, or may represent a series of sample values
from lab analysis. Suppose the analysis is of an interme-
diate process stream that goes to a mixing tank. Then
tank variations will be attenuated to represent low fre-
quency swings in the final product where quality varia-
tions have a meaningful economic influence. For the given
period oftime, quality measurement is variable and has an
average value within the spec limit. Values only approach
or violate the spec on rare occasions. The improved control
system will reduce these fluctuations in the product qual-
ity so that the average value may be moved closer to the
spec without violating it. This is invariably the most prof-
itable direction in which to move product quality.
For example, in blending a low sulfur fuel oil, the closer
the sulfur specs are approached, the less the requirement
Table 2. Suppose the spec is not met
S > spec
profit = selling price - cost
P = SP - C, e/lb
Case 1. Worthless: SP = 0, so P < 0
Case 2. Penalty: 2% S worth 14.55 $/b
1% S worth 15.55 $/b
1OO¢/b/%S = 0.333¢/lb/%S
SP = SPa - 0.333 (S - spec)
P = SPa - u (S - spec) + (C2 _ C1) [ S - S2 ] - C2
S1- S2
Case 3. Lose customers: C1 increases at lower volume
C1 = C1a + ~(S - spec), ~ > 0
C = (C1a - C2) [ :1 ~ ~~ ] + C2 + ~(S - spec) [ :1 ~ ~~ ]
[
S - S2]
P=SP+(C2-C1a) --- -C2
S1- S2
~ .____ (S2 -S(S2+spec)+S2spec)
S1-S2
for the most valuable low sulfur blending components. As
gasoline octane approaches a minimum spec, the gasoline
yield or operating cost for adding lead anti-knock com-
pounds becomes more favorable. As the boiling range of
heavy naphtha from a crude distillation tower approaches
its maximum limit, the yield of more valuable gasoline
increases. As fuel oil viscosity approaches a maximum,
its yield or operating cost is more favorable. As the tem-
perature on a crude unit heater approaches an upper limit,
yield of the most valuable products increases. As the flood-
ing point on a distillation column is approached, column
capacity increases. As compressor speed approaches its
limit, plant capacity increases. As average lube can fill
approaches label content, product loss is reduced.
Therefore, the normal procedure for justifying invest-
ments in improved process control is to estimate economic
improvement by moving the average value closer to spec.
This is a steady-state concept that really relates to chang-
ing from the second curve to the third curve in Fig. 1. The
process control system is usually viewed as a device that
will allow that second step by shrinking transient varia-
tions from the first to the second curve. In fact, no eco-
nomic benefits are credited for that first step alone, based
on improved dynamic performance. Only the second step
of moving the average value closer to the spec gives an
economic benefit. This benefit is determined from plant
engineering models for yields, capacity and operating
costs. The first step is a necessary prerequisite.
This concept is often viewed in terms of distribution
functions for data collected over a period of time? as indi-
cated in Fig. 2. The upper curve represents distribution
of quality points about the average value or target value.
It indicates that a small fraction of the data violates the
spec limit. (If the limit is never violated, the average or
target is too conservative.) The improved control case
shows data points clustered more closely about the aver-
age value that has been moved closer to the spec. If the
standard deviation, 0, is reduced by a certain amount, the
difference, 11, between the limit and the target average
can thus be reduced by a corresponding amount.
This procedure of assessing improvement in the steady-
state average quality and assigning an economic benefit to
that improvement is probably the easiest way to under-
stand improvements from process control. However, it stands

Problem description. The basic prob-
lem is to determine a clear and direct
way to relate dynamic criteria to pro-
cess economics. Consider two separate
steps: The first is to reduce variations
alone by improved regulation about the
same average value. Several questions
then arise: What is the net benefit of
this step alone? Is it zero? Do the fluc-
tuations really cancel out? The second
step is to move the average value closer
to the spec and determine net benefit of
this step alone. The difficult problem is
evaluating the first step.
There is always economic value as so- Fig. 5. Profit depends on quality.
ciated with step one alone. But estab-
lishing this benefit is often obscure and difficult because it
requires modeling the penalty for violating the spec. These
models are outside the domain of plant modeling. They
include market customer satisfaction, legal effects, safety,
maintenance, human creativity and the environment.
These sources of hidden benefits of improved quality con-
trol are invariably overlooked. These hidden benefits are
often very significant.
Distribution curves for step one are illustrated in
Fig. 3. In the improved control case, the standard deviation
has been reduced. Product is more uniform, but its aver-
age product quality is unchanged. Does the plant or cor-
poration derive any economic benefit by better product
uniformity if the average value is not moved closer to the
specification? Yes!
First, the amount of off-spec product in the distribution
tail is reduced by either increasing the margin, 6, away
from the spec or by reducing the standard deviation, 0, as
occurs in the control case. It is clear that if a model could be
derived for the economic penalty that resulted from the
spec violations in the base case, one could determine the
improvements by virtue of the fact that fewer samples vio-
lated the spec after control was improved.
In the base case with a fixed 01, consider the problem
of plant management in specifying the best average tar-
get somewhat within spec limit. Certainly one can imag-
ine that ifthe target is far below the limit there will be an
important economic loss. But as the target average is
moved closer to spec limit, the fraction of off-spec mate-
rial produced begins to increase. Clearly this will have an
economic impact on the long-term plant performance.
This concept is built into the common sense statement,
"It's better to operate on the safe side." What this means
is that the true economic optimum point for the target
average is at some point that is offset from the spec by 61,
to trade off yield benefits with customer dissatisfaction.
in sharp contrast to lessons ofthe tech-
nology of optimal dynamic control the-
ory developed since 1960. Control theory
teaches us how to design feedback con-
trol systems that will optimize dynamic
performance of the plant with respect to
a certain dynamic performance criteria.
We know we must model the economic
impact from violating the spec.
Example: fuel oil blending. Heavy
no. 6 fuel oil is a major refinery prod-
uct typically burned in boilers for
steam and electricity. The sulfur con-
tent spec is typically less than 1% to
ensure satisfactory S02 emissions
from boiler to atmosphere. This fuel
oil product is usually made by blend-
ing two components (Fig. 4). One is a low sulfur fuel oil
(LSFO) slightly above the sulfur spec. It is upgraded with
a more valuable low sulfur light gas oil (LGO). The mini-
mum amount of valuable LGO will be used when the final
product is at its sulfur spec. If the product has less than
this sulfur content, too much valuable LGO was used with
a resulting economic loss from quality giveaway. For given
prices for the blending components, LGO and LSFO and
their sulfur content, a weight balance blending equation
gives the cost for producing no. 6 fuel oil (Fig. 4).
This equation is rewritten in general terms of profit
at the top of Fig. 5. The profit equation is plotted in
Fig. 5 in terms of the quality ofthe product's sulfur con-
tent. This steady-state equation applies for perfect qual-
ity control. As the sulfur is increased, profit increases lin-
early. Maximum profit occurs when the product is at spec.
A simple calculation at the bottom of Fig. 5 indicates eco-
nomic improvement by moving average sulfur content
from 0.95 to 0.98. The slope of this curve is easy to deter-
mine from knowledge about the process. One can imagine
similar process relationships for distillation columns,
chemical reactors, and any manufacturing operation.
We will continue to assume perfect control in steady-
state concepts. Now consider what happens when a spec is
not made (Table 3). The profit still remains selling price
minus manufacturing cost. But now we must consider the
selling price part of the equation rather than the cost part.
Look at three situations:
In one, the product is worthless when the spec is vio-
lated. The selling price would be zero. Profit becomes neg-
ative. This represents a severe discontinuity in the profit
equation of Fig. 5.
In the second situation, there might be a sliding penalty
on selling price based on actual sulfur content ofthe sam-
ple. This is typically the case for no. 6 fuel oil, shipped by
tanker to the U.S. East Coast. Upon delivery it may be
Profit = selling price - cost
P = 5P - C, e/lb
[s-S2JP = SP + (C2 - Cl) ~ - C2, C2 > Cl
or-__-.~--------~Q-u-a-li~~,~S~
Spec
Example calculation: average sulfur
from 0.95 to 0.98
AP - (C2 - Cl) [0.98 - S2] -(C2 - Cl)
- Sl - S2
[
0.95 - S2J= (C2 _Cl) [0.98 - 0.95J=
Sl-S2 Sl-S2
(6.65 - 5) [ 1.10~0~.2J
AP = 0.055 ¢/Ib = 16.5 ¢/b
Cl = Cost of LSFO, ¢/lb
C2 = Cost of LGO,¢/lb
51 = LSFO sulfur, wt %
S2 = LGOsulfur, wt %
;;::::
e
~Or---~------~--+-~--~
P = SPa - a(S - spec) +
(C2 -Cl) [S - S2J - C2
51-52
Fig. 6. Spec violation penalty.

sampled by the customer before acceptance. For example,
if2% sulfur has one lower price and 1% sulfur has another
higher price, a sliding scale for the price per pound per
percent sulfur can be established. Hence, actual selling
price is given by the equation with the sliding penalty
being a function of how much the analysis exceeds spec.
The actual profit equation expended to include this term
is in Table 3. These two terms for perfect control are plot-
ted in Fig. 6. The penalty term for reduced selling price
Table 3. Fuel oil blend examples
Example 1:
profit = selling price - operatinq cost - loss
p= SP - Co - K1(spec - s) -K2R
where K1 = ¢/Ib/% sulf. = 0.333
R = fraction rejected. Depends on t!Jcr from cumulative
probability curve
K2 = c/lb cost incurred by violating spec. reprocessing cost
Example 2:
spec = 1% ~ulf.
Base case SOPT = 0.95%
cr1 = 0.99 - 0.95 = 0.04
cr2 = 0.97 - 0.95 = 0.02
Ll.1 = 0.95 -1.0 = -0.05
~= 0.05 =-1.25
cr1 0.04
Cost = K ~ _ K !i
o 'cr 20
At optimum
d cos t / 0 = K _ ~ ~ I = 0
dLl./o ' 0 dLl./o 0
dR IdLl./o 0
K,o
K2
Example 3:
At base optimum Ll.1/01 = -1.25,
from normal probability curve
dR K,01
--=0.1827,K2
dLl./o 0.1827
At new optimum
dR K cr2 K cr2
--=-' -=-' -(0.1827)=0.0913
dLl./cr K2 K,01
From normal probability curve
Ll.2=-1.715
02
so
Ll.2= -1.715(0.02) = -0.0343
New SOPT = 1 - 0.0343 = 0.9657%
Base case SOPT = 0.9500%
R1 = 0.1056, R2 = 0.0431
Example 4:
Increased profit is:
P2-P1 = Ll.P=K1 (Ll.2-Ll.1)-K2(R2-R1)
Ll.P = K1 (-0.0343 + 0.05) + K2 (R1 - RiJ
= 0.0157 K1 + K1 cr, (R1 - RiJ/0.1827
From normal probability curve:
Ll.P = 0.0157 K, + 0.0137 K, = 0.0294 K,
profit improvement = steady-state term + fluctuation penalty term
(53%) + (47%)
Hidden benefit = 47 = 89% of steady- state benefit ~
53
This profit
benefit, ~P2,
is a real
contributorl
to company
profits.
Where does
the money
come from?
could be severe, and net profit
drops rapidly. Of course, the
maximum profit under perfect
control is still achieved by oper-
ating precisely at spec. Qual-
ity/price data are readily avail-
able from public trading
information such as the New
York Mercantile Exchange.
In a third situation, repeated
violations cause the plant to lose
customers. As production volume
decreases, some operating costs
increase. For example, Cl may
increase as indicated by the equa-
tion in Table 2. In this situation
the profit equation becomes more
complex as a quadratic function
of quality. These situations rep-
resent models ofthe market and
economic environment in which the plant is operating. The
best source of information for these models is legal con-
tracts for product sales to customers and marketing studies.
Economic models for reputation, reliability, speed of
response, customer satisfaction, sales costs,advertising
effectiveness, discounts, refunds, legal penalties, insurance
ratings, fraud and neglect are empirical, volatile and com-
plex. So they usually remain intangible, but very real.
Hidden benefits from reduced variations. Returning
to Fig. 6, we know that in practice there is uncertainty in
the measurement and control of product quality, and the
average target should be at some safe value below spec. A
realistic situation accounting for transients is in Fig. 7.
This is a plot of actual profit versus average or target qual-
ity. Given the variability or distribution function in the
base case, we know there must be some economic penalty
because the average is moved closer to the limit, and the
tail begins to exceed the limit.
In reality there is an optimum profit point, as illus-
trated, that is within spec. When experience says it is
"better to operate on the safe side," we really mean that
the maximum profit point does not occur at spec but at
a point within spec. This is because the economic penalty
for violating the spec begins to become significant and
affect the overall performance curve before the average
value reaches the limit, as illustrated. The significance
and position ofthe penalty term in Fig. 7 depends on the
uniformity of product quality, oI. This curve describes
operation's profit as a function of adjusting the average
quality target value closer to spec under a fixed control
system performance with a fixed distribution curve and
standard deviation, ol.
The improved control case is in Fig. 8. Here, variations
have been reduced, and the average value can be safely
moved closer to the true spec. The optimum point is as
illustrated because the penalty term for violating the spec
does not become significant to the profit curve until the
average is closer to the spec point. Notice that this is a
plot of profit versus the average quality or.target value,
with a fixed control system performance, or a fixed cr2,
less than o I.

These two curves are superimposed
in Fig. 9. Note that the curve for the
improved control case has a different
shape and a different long-term steady-
state profit function. Note also that the
actual profit improvement from the
best point on the base curve to the best
point on the improved control curve is
the sum of two components, ~1 and
tJ>2. The first component, tJ>1, is deter-
mined from the slope of the cost por- Fig. 7. Profit depends on average quality.
tion of the curve and the difference
between the base optimum target value
and the improved optimum target
value. ~1 represents a steady-state
cost benefit from moving the average
quality closer to the specification: Step
2 in Fig. 1. This term is the most easily
computed and is commonly used for
justification. The second term, ~2, is
the fluctuation penalty benefit that is
derived from more uniform quality
alone at the same base optimum aver-
age point. This is the hidden compo- Fig. 8. Profit depends on average quality.
nent that is hard to quantify and is
invariably overlooked, because it depends on the penalty
model for violating the spec.
This profit benefit, tJ>2, is a real contributor to company
profits. Where does the money come from? This benefit
money must come from the fact that there are fewer viola-
tions of the spec. If one argues that in the base case there
is no violation ofthe spec, one is saying that in the base case
the operation is far to the left ofthe base case optimum point.
This means an additional benefit is realized from moving
up from that point to the base optimum point by an operat-
ing cost reduction with the existing control system. If the
spec is never violated, the target is not optimum. Since this
can be done immediately and at no cost, it should be done,
and the control system base case must be at the optimum
point in Fig. 7.
The situation in Fig. 9 is a conservative evaluation
because it assumes that the base case operation is indeed
operating at its optimum point. The definition of this
smooth optimum is the point of trade-off between the cost
curve and the economic penalty curve for violating the
spec. Since the model of the spec violation curve is so dif-
ficult to obtain, it is usually neglected, and the component
~2 is thereby also neglected.
Another observation in Fig. 9 is that as the standard
deviation is reduced or the control fluctuations are
reduced, the shape ofthe steady long-term profit function
changes. At constant average quality, the profit increases
by ~2 from the lower point to the middle point.
Base case
~o~---,~--~~~--~--~~
0:

Spec Penalty
Improved
control
Fuel oil blending case. An example problem was solved
in detail to estimate the relative magnitude ofthe two profit
terms, tJ>1 and tJ>2. The summary results ofthis example
for the fuel oil blending case are given in this section.
The complete profit function is given in Table 3. The
last loss term is a constant, K2 multiplied by R, the frac-
tion of the total production that is rejected by violating
the spec. Assuming the actual product quality distribu-
tion is a random Gaussian distribution, the fraction
rejected depends on the ratio for the margin, !:J., to the
AP1
.0
s
~or---~------~~~~--~~
0:
Base
Spec case
Improved
control
AP1 = Steady-state cost benefit from
moving average quality closer
to specification
AP2 = Fluctuation penalty benefit from
more uniform product quality
alone
Fig. 9. Improved control has a different profit
function.
Table 4. Practical applications
1. Fuel oil sulfur-less complaints/reprocessing
2. Gasoline octane-less lost customers
3. Naphtha endpoint-longer reformer catalyst life
4. Fuel oil viscosity-less complaints/reprocessing
5. Furnace temperature-less coking, longer tube life
6. Distillation flooding-less faulty operation
7. Compressor rpm-Iess compressor maintenance
8. Gas oil 90% point-less metals in FCC feed
9. Lube oil can-less liability for fraud
Table 5. Theoretical implications
1. Optimal control theory: based on linear mathematics
Legal specifications: based on nonlinear mathematics
2. Realistic dynamic perfonnance criteria can be related to steady-
state dollars
3. Important model requirement is the economic debit for violating
specifications
deviation, cr. In this model we assumed that any prod-
uct that violates the spec, no matter how severely, incurs
a fixed cost per pound for reprocessing.
A specific case is given in examples 2 and 3 of
Table 3. Here the base case, ol , is 0.04, and the control
case reduces this by half to 0.02. Using the cumulative
normal probability curve, the equivalent margin on the
control case is found to be - 0.0343, which means the
new optimum average target value can be moved up from
0.95% to 0.9657%. Using the profit equation, the difference
between the two profit levels is given in example 4 of
Table 3 by subtraction. The first term arises from a reduc-
tion in cost, and the second term arises from a reduction
in the fraction of product rejected.
Notice that the fraction rejected in the control case is
less than the fraction rejected in the base case, in spite of
the fact that the average value is closer to spec. From the
normal probability curve, we find these two terms given in
example 4 of Table 3. The steady-state term contributes
53% of the total profit for improvement, and the fluctuation
penalty term contributes 47% to total profit improvement.
Therefore, the fluctuation penalty improvement is 89%

of the steady-state term that is normally used to assess the
benefits. This means that in this case, benefits from the
improved control system are almost twice as great as
would be estimated by the usual approach. Although spe-
cific results for other cases depend on the model for vio-
lating the spec and the particular numerical values, the
general symmetry of the optimum profit curves leads us
to believe that this is a fairly typical result. In fact, when
spec violation penalties are much more severe than cost
saving credits, the hidden benefit exceeds the steady-
state process term.
Results. Experienced users of advanced control systems
and computerized real-time plantwide information systems
with SQC, SPC and just-in-time (JIT) quality technology
are realizing and reporting substantial results from the
intangibles. Champlin Refining and Petrochemicalsf
recently documented tangible benefits of$10 million a year,
with intangible benefits,
since 1990, of a similar
amount. With refiners
facing up to 10 govern-
ment-imposed quality
specs on future reformu-
lated fuels, and U.S.
investments exceeding
$4 billion a year for the
1990s to meet environ-
mental standards, the
performance andjustifi-
cation of quality control
and information systems
is gaining serious atten-
tion. The experience of
Japan during the 1980s
has already provided tes-
timony on the existence
and magnitude ofintan-
gible hidden benefits
from quality control."
We are left simply
with a remodeling problem for the value of improved per-
formance, with plant models of physics and chemistry
only half of the story. Models of the external impacts are
also needed. Once the complete penalty model is deter-
mined, benefits follow. One ultimate benefit may be cor-
porate survival.
One should investigate the consequences of violating
a specification, and quantify it. Reduced fluctuations alone
are beneficial. The improved control case has a different
profit function. The benefit from improved SQC/SPC is
greater than conventionally estimated, probably double.
One should
investigate the
consequences
of violating
a specification,
and quantify it.
Reduced
fluctuations alone
are beneficial
Commercial applications. The results of this study
have many commercial applications for petroleum refining.
Some are listed in Table 4. Many real benefits from im-
proved process control are often lost in the intangible cat-
egory because they are so hard to quantify. In the fuel oil
blending example, the additional benefits come from fewer
complaints or less reprocessing of off-spec material. In
blending gasoline to meet octane specifications, the hidden
benefit comes from fewer lost customers or better cus-
tomer satisfaction and increased sales. From improved
control ofnaphtha/kerosine cutpoint on crude distillation
Copyright<>1992 by Gulf Publishing Company. All rights reserved.
units, the hidden benefit comes from longer catalytic
reformer catalyst life. The other spec hidden benefits are
self-evident. Each ofthese could be the basis for indepen-
dent study using this approach. Major modeling work is
justified for the effects of pressure, temperature and com-
position on equipment wear and tear, safety hazards, cor-
rosion, plugging, fouling, ruptures, run length and main-
tenance costs. This know-how is available but rarely
quantified for SPC benefits.
There are some important theoretical implications to
process control as a result of this study (Table 5). Most
of the significant results in optimal control theory that
lead to closed form solutions for control law algorithms
are based on linear process models and linear mathe-
matics. In contrast, most real world performance crite-
ria are encoded in legal specifications that are strongly
nonlinear. The second point is that realistic dynamic per-
formance criteria can be related to steady-state dollar
benefits. The third point is that a very important model-
ing work requirement determines the economic benefit
for violating quality specs. Process control engineers, by
training and interest, emphasize modeling the physics
and chemistry of the process plant to determine the effects
on yield, quality and energy. But they usually overlook
additional performance modeling work that is not related
to chemistry or physics as much as plant economics and
market performance factors for products.
More study is needed on the relationship and impact
of different types of economic spec models on the design
and performance of process control systems. Another impli-
cation is that the optimum average target value depends
on variability, which is affected by control system perfor-
mance in addition to spec value. Another result is improved
dynamic quality control performance benefits come from
both process yield and customer satisfaction, and not a
trade-off of one sacrificed for the other.
LITERATURE CITED
1 Shinsky, F. G., "The Values of Process Control," Oil & Gas Journal, February 18, 1974,
p.80.
z Taylor Instrument Co., Refinery Process Control, Rochester, N.Y., p. 2-l.
3 Jones, C. A., "Review and Evaluation of Philadelphia Refinery Computer Control Sys-
tem," NPRA Computer Conference, Chicago, November 15, 1973.
4 Latour, P. R., "On-line Computer Optimization 2: Benefits and Implementation," Hydro-
carbon Processing, Vol. 58, No.7, July 1979, p. 219.
5 Latour, P. R., "Economics of Advanced Control and Digital Systems," lecture at first bi-
annual short course on Application of Advanced Control in the Chemical Process Indus-
tries, University of Maryland, College Park, Md., May 12, 1981.
6 Martin, G. D., Turpin, L. E. and Cline, R. P., "Estimating Control Function Benefits,"
Hydrocarbon Processing, Vol. 70, No.6, June 1991, p. 68.
7 Sharpe, J. H. and Latour, P. R., "Calculating Real Dollar Savings from Improved
Dynamic Control,"Texas A&M University Annual Instruments and Control Symposium,
January 23, 1986.
8 Thompson, w., Catt, M. L. and IGpper,J. P., "Benefits ofaRefinery Information and Con-
trol System," Paper cc-91-140, NPRA Computer Conference, Houston, Texas, ovem-
ber 13, 1991.
9 Latour, P. R., "Advanced Computer Control of Oil Refineries-Where We Are, Where We
Are Going," Paper 21, Petroleum Refining Conference, Tokyo, The Japan Petroleum Insti-
tute, October 21, 1988.
The author
Pierre R. Latour is vice president of marketing
and business development for SETPOINT, Inc.,
Houston. Previously, he directed central market-
ing and the oil refining advanced process control
division. Dr. Latour was founding chairman of
the board of SETPOINT Japan, Inc. He is a co-
founder of both SETPOINT and Biles & Associates
in Houston. Prior to joining SETPOINT, he was a
senior engineer for Shell Oil, Deer Park, Texas, and Shell's head office
in New York. Dr. Latour holds a BS degree in chemical engineering
from Virginia Tech and MS and PhD degrees in chemical engineer-
ing from Purdue. He is also the author of numerous technical papers
on advanced process control and optimization.
FI1M/692 - Printed in United States

QuantifyQC HPMay92

More Related Content

What's hot (19)

Similar to QuantifyQC HPMay92 (20)

More from Pierre Latour (20)

QuantifyQC HPMay92