Vous êtes sur la page 1sur 6

Reactions and Separations

Estimate the Capacity of


Simple Batch Processes

Finding the throughput of a multiple-unit,


multiple-step operation, even when it is
James L. Manganaro, relatively uncomplicated, is not always an easy
Consultant matter. Several equations and graphical aids
are provided to help determine yearly capacity.

T o the chemical engineer who has worked


with large-scale processes that are generally
run in a comparatively orderly continuous
fashion, the analysis of a batch-processing
problem can be new and somewhat mysterious.
Batch operations often arise in small-scale produc-
tion of high-value products. The batch process consists
a bottleneck. Since the units in each step are usually
of incremental capacity, it may not be possible for
the designer to exactly match the equivalent cycle
time of each step. Moreover, the units within a step
may be operating at different cycle times. A contrib-
utor to the non-uniformity of the equipment roster is
the desire to produce variations in product character-
of a series of steps in sequence. A batch, or several istics that may call for different units (with different
batches, is prepared in the first step over a period of cycle times) within a step. Also, identical units
time. The batch(es) from the first step is then delivered might be run under differing conditions, with conse-
to the second step, where the units (the different pieces quently differing batch times.
of equipment) carry out further operations, and so on. Further, once batchwise production is in place,
Each step may be composed of several identical or capacity expansion is accomplished by the simple
often-different units, which yield the desired result. For expedient of adding more units to each of the steps.
example, a reaction step may employ two different Since this is often done over many years, the equip-
kinds of reactors, such as a fluidized bed and a rotary ment that are added to each step may not be identical
kiln, each having a different cycle times and possibly to those already in place.
producing products of differing properties. Much of the literature (13) on batch process de-
If a particular step is not the bottleneck in the se- sign and operation focuses on accommodating: (a)
quence, then some of the units within the step might be the production of a plurality of different products;
idle during operation. By idle, we mean that the func- (b) a fluctuating customer demand; (c) raw material
tion of the unit has been accomplished and that that availability and price; and (d) a plants equipment
piece of equipment is either empty and waiting for feed status. Sophisticated, comprehensive software exists
from upstream or that it is full, but waiting for some for the analysis and operation of batch processes of
downstream equipment to first be unloaded. If a partic- this order of complexity. However, the engineer may
ular step is the bottleneck, then the equipment within not have access to or require the full reach of these
that step may not (or may) see idle time. tools. Therefore, this article considers several rela-
Good process design calls for matching step cycle tively limited, but nevertheless realistic, problems
time, or what we will define as the equivalent and presents approximate, but simple, ways of esti-
cycle time of each step, so that no one step becomes mating the capacity of a batch process.

70 www.cepmagazine.org August 2002 CEP


We will examine a process that consists
Table 1. Example batch process with units of equal cycle time in a step.
of a sequence of batch steps in which each
is composed of a number of units that can Step Number of Cycle Time of Each Equivalent Cycle
be of differing cycle times. We will sug- Units in Step Unit in Step, h Time of Step, h
gest a simple means of estimating: A 2 2 1
1. the equivalent cycle time of a step B 4 4 1
2. the bottleneck step from Number 1 C 3 3 1
3. an upper bound of capacity of the
overall batch operation from Number 2
4. the fraction of product produced by a given unit of Lower-case c is used to denote cycle time of a unit with-
the bottleneck step. in a step, while upper-case C denotes the equivalent cycle
In the following discussions it is assumed that batches time of a step, and the asterisk signifies the special case of a
are all the same size and there is 100% availability of step with units of equal cycle times. Cycle time is often given
equipment, unless otherwise stated. It is also assumed that as hours, when it is tacitly understood to be hours per batch.
any reagents needed for intermediate steps are always Determining the equivalent cycle time of a step that has units
available. We will distinguish between process cycle time, of unequal cycle times will be discussed later.
step-equivalent cycle time and cycle time of a unit. The last
is self-explanatory. Processes having steps composed of units
Process cycle time The overall process cycle time, of equal cycle times
CP, is the time per batch of final product produced calcu- Here, we offer two examples of batch processes, each
lated over a period of time and a large number of batches. consisting of three steps and having steps that are com-
The process cycle time defines the capacity of the pro- posed of units of equal cycle times. In the first case, the
cess, see Eq. 8. batch process has steps of equal equivalent cycle times,
Equivalent cycle time of a step The equivalent cycle while in the second case, one step has a longer equivalent
time of a step, Ci, may be defined as the time per batch for cycle time than the other two steps.
the step calculated over a period of time and a large num- A process with steps of equal equivalent cycle time
ber of batches and assuming that feed is always available Consider a batch process consisting of three steps, such as
to the step being considered. batch preparation, reaction and filtration. Say that each of
Limiting step of a batch process The limiting or bottle- the units of a given step has the same cycle time and that
neck step of a batch process is defined as that step having the equivalent cycle time of each step, as defined by Eq. 2,
the greatest equivalent cycle time. Thus, the equivalent cycle are equal. Specifically, take the example given in Table 1.
time of the limiting step is given by: Applying Eq. 2, the equivalent cycle time of each of the
steps is 1 h. Such a process can be represented by Figure 1.
Cmax = max{Ci}all i (1) In this figure, there is perfect dovetailing of each step with
the subsequent step; no unit of equipment is ever idle.
In a continuous process, the limiting step has Also, each of the steps produces a finished batch ready for
sovereign control over the capacity of the entire process. the next step every hour, i.e., with a 1-h process cycle time.
However, in a batch process, while the limiting step This special case suggests that if the units within each step
often exclusively controls the capacity of the overall pro- have identical cycle times, and if the equivalent cycle times
cess, there can be circumstances in which improving the (as defined by Eq. 2) of all the steps are matched, then the
cycle time of the next limiting step can also increase the
overall process capacity.
P
Equivalent cycle time of a step composed of P
A1 B1 C1
units of identical cycle time P
A2 B2 C2
To begin, it is initially assumed that the units in a given P
step are of identical cycle time, c*. Then the equivalent A1 B3 C3
P
cycle time for the step is the cycle time for a single unit di- A2 B4 C1
P
vided by the number of units in the step. Thus, for this spe- A1 B1 C2
cial case, the equivalent cycle time of the step is: A2 B2 C3
Time

c*
C* = (2)
Figure 1. A batch process with three steps of equal equivalent
N
cycle times.

CEP August 2002 www.cepmagazine.org 71


Reactions and Separations

Equivalent Cycle Time of Step


P

Normalized to Faster Unit


P 1.0
A1 B1 C1 0.9
P P
A2 B2 C2
0.8
A1 B3 C1
P P 0.7
A2 B4 C2
A1 B1 C1 0.6
P P
A2 B2 C2 0.5
A1 B3 C1 0 5 10 15 20 25
A2 B4 C2 Cycle Time of Slower Unit Normalized to Faster Unit
Time

Figure 3. Normalized equivalent step cycle time for two units of unequal
Figure 2. A batch process with three steps of unequal equivalent cycle- cycle times.
times. This is similar to Figure 1, except that Step A has two units of 3 h
cycle time.
rather than of 1.0 h. Then, step A would have the longest
equivalent cycle time (compared to the 1.0 h of steps B and
process will have a capacity simply calculated from the C), and would be regarded as the limiting step. A represen-
equivalent cycle time. This, as will be seen, is not quite tation of this case is shown in Figure 2. Note that the units
true for batch processes where the steps are comprised of of step A are always in operation, while units B1 and B2 of
units of unequal cycle times. step B are sometimes idle, and unit C3 is always idle. The
A process with steps of unequal equivalent cycle times cycle time per product batch (i.e., the process cycle time) is
Now consider the case given in Table 1, but in which now 1.5 h, which is, in fact, the equivalent cycle time of
the cycle time of the units in step A is 3 h, instead of 2 h. step A. Therefore, the production rate of this process can
This would give step A an equivalent cycle time of 1.5 h be computed from the cycle time of the limiting step.

Steps composed of units of unequal


Nomenclature cycle times
c = cycle-time of a unit in a step in which the units have unequal Consider a step made up of two units, but the units have
cycle-times, h/batch unequal cycle times. Again, we assume that feed is always
c* = cycle-time of a unit in step in which all other units in this step available to this step, so that neither unit is ever idle. If the
have same cycle-time, h/batch units were of equal cycle time, then the equivalent step
cmax,j = cycle-time of unit j of the limiting step, h/batch
cycle time would be one-half of the cycle time of either
C = equivalent cycle-time of a step where the units have unequal
cycle-times, h/batch unit. If, at the other extreme, one of the units had an in-
CP = overall process cycle-time, h/batch finitely long cycle time, then the equivalent step cycle time
C* = equivalent cycle-time of a step whose units have identical would be just the cycle time of the faster unit. Thus, the
cycle-times, h/batch equivalent step cycle time normalized to the faster unit
Cmax = equivalent cycle-time of limiting step (i.e., maximum should lie between 0.51.0.
equivalent cycle-time of steps), h/batch Determining the equivalent step cycle time for this case
fo = onstream factor, fraction in which feed to the step is always available corresponds to
fs = scheduling factor, fraction
high-school algebra word problems such as: Sammy can
Fmax,j = fraction of total product produced by unit j of limiting step
N = number of units in step in which all units have same cycle-time shovel the walk in 2 h while Billy (several years his junior)
Ni = number of units in Step i can shovel it in 5 h. How fast can the two of them working
P = annual capacity, million lb/yr together remove the snow?
w = weight of product per batch, lb/batch Therefore, the step equivalent cycle time for step i is re-
lated to the cycle time of the shorter, cshort, and longer, clong,
Subscripts: units by:
i = Step i
j = unit j of Step i
1 1 1
Superscript: = + (3)
* = units within the step have identical cycle-times Ci cshort clong

Eq. 3 can be recast so that equivalent step cycle

72 www.cepmagazine.org August 2002 CEP


Figure 4. Two-step batch process
in which the Step B has units of P
unequal cycle-time. P P
A B1
A B2 P
A B1 P P
A B1
A B2 P
A B1 P
6h P
A B1
10 h A B2
A B1
Time 14 h
18 h
22 h

time and long cycle time are normalized to the shorter product that is produced by the faster unit is 87.5%.
cycle time, see Eq. 4: Of course, the generalization for a step with any num-
ber of units having unequal cycle times and operating
clong under the assumption that feed to the step is always avail-
Ci cshort able is clear:
= (4)
cshort clong
1+ 1
cshort Ci = Ni (5)
1
Eq. 4 is plotted in Figure 3. The normalized step equiva-
j =1 ci , j
lent cycle time for this case varies between 0.5 and 1.0, as it
should. The ratio Ci/cshort is, in fact, also the fraction of total If the cycle times of the units in the step are equal, then
product that is produced by the faster unit. Eq. 5 reduces to Eq. 2.
To take a numerical example, consider the shorter
cycle unit of the step having a cycle time of, say, 4 h and Example of a two-step batch process
the longer cycle unit, of 28 h. Then, the normalized In a two-step sequential batch process, just determining
cycle time of the slower unit is 28/4 = 7, and, therefore, which of the steps has the the longer equivalent cycle time
the normalized equivalent step cycle time from the and then basing the capacity calculation on this supposed
graph is 0.875, so that the actual equivalent step cycle bottleneck step is, unfortunately as will be seen, not the
time would be 0.875 4 = 3.5 h. Also, the percentage of complete picture. In other words, the overall process cycle
time, CP, is not necessarily equal to but may be greater
than the equivalent cycle time of the limiting step, i.e.,
Table 2. Two-step batch process with Step B having
units of unequal cycle times.

Step Unit Cycle Time of Step Equivalent


Unit, h Cycle Time, h
6
A A 2 2 Step A has only 1 unit that varies as indicated.
Process Cycle Time, h

5 Step B has 2 units: B1=4 h, B2=6 h cycle time.


B B1 4 2.4 (by Eq. 5) Equivalent cycle time of step B is 2.4 h
B B2 6 4
3
2 Equivalent cycle
Table 3. Two other examples of Step B having an 1 time of Step B
equivalent cycle time of 2.4 h.
0
Sub-case B1 Cycle B1 Cycle Step B Equivalent 0 1 2 3 4 5
Time, h Time, h Cycle Time, h
Step A Cycle Time, h
1 4.8 4.8 2.4
2 2.5 60 2.4

Figure 5. Two-step process cycle-time yields some unexpected results.

CEP August 2002 www.cepmagazine.org 73


Reactions and Separations

CP Cmax (6) limiting step, then reducing the cycle time of the next
limiting step could reduce the cycle time of the overall
Consider the two-step batch process described in process.
Table 2. The equivalent step cycle time of step A is 2 h Now consider two sub-cases to the one in Figure 5, each
and that for step B, by Eq. 5, is 2.4 h. Consequently, one having a step B equivalent cycle time of 2.4 h (Table 3).
would be inclined to say that the capacity of this process In sub-case 1, both units B1 and B2 have equal cycle
could be calculated using only the equivalent cycle time times of 4.8 h, giving a step B an equivalent cycle time
of step B, since it is the bottleneck step. However, if we of 2.4 h. In this sub-case, the ideal dashed curve in Fig-
examine a representation of this batch process (Figure ure 5 would have been followed. In sub-case 2, where
4), it is seen that the assumption that feed is always the cycle times of the two units of step B are widely dif-
available to step B is not true. ferent, it was found that the curve for this case (not plot-
Figure 4 shows that at the end of 8 h, step A is ready to ted) is closer than the original case of Table 2 to the
deliver its finished goods, but neither unit B1 nor B2 is ideal dashed curve.
ready. At the end of 10 h, both B1 and B2 are ready to re- It appears that there is a region of mild disparity (neither
ceive feed, but step A can deliver to only one of the units, equal nor vastly different, as in Table 2) between the cycle
B1. The other unit, B2, has to wait another 2 h before feed is times of the units in step B that results in the greatest wait-
ready for it. ing time for the limiting step and, therefore, the widest di-
Thus, the assumption of feed always available to step B is vergence from the ideal dashed curve in Figure 5.
not true for this set of unit cycle times. This bleak state of af- Since only a simple two-step process has been consid-
fairs can, of course, be partially remedied by the use of surge ered here, it is not clear what happens in the case of multi-
capacity between steps A and B. ple steps composed of units of unequal cycle times. Do the
The repeating pattern in the description illustrated by non-idealities accumulate?
Figure 4 shows that three batches are produced every 8 h. We can also ask what happens when step A in Figure 5
Thus, the equivalent cycle time for the yoked steps of A has a cycle time of 2.4 h and we slightly reduce the cycle
and B, as shown in Figure 4, is (8/3) = 2.67 h/batch; not times of either unit B1 or B2 from the original values of 4 h
the 2.4 h//batch given by Eq. 5, which was based on feed and 6 h, respectively. It is found that a small reduction in
being always available to the so-called limiting step. the cycle time of the faster unit, B1, does not have any ef-
We can examine how the overall process cycle time of fect on process cycle time, while a small reduction in the
this two-step process varies with changes in the cycle cycle time of the slower unit, B2, is indeed reflected as a
time of step A, holding step B fixed at the cycle times in- reduction in process cycle time.
dicated in Table 2. This was done by constructing a True The True Basic simulation indicates that, if the limit-
Basic computer program that simulated the actual ing step has waiting time for feed, then in some, but not
scheduling, as for example, as shown in Figure 4. The all, cases, giving preference to the shorter cycle time unit
computed effect of step As cycle time on the overall pro- of the step will result in a shorter process cycle time. If
cess cycle time is shown in Figure 5 (see the line with there is no waiting time for the limiting step, then, clear-
the solid-circle markers). ly, giving preference to the short cycle time unit of the
The dashed curve is what would be expected if the step does not matter.
process cycle time were simply always equal to the lim- Fraction of product produced by a unit in the limiting
iting step equivalent cycle time. This is the ideal case, step When the units in the limiting step are different and
in which the units in the limiting step are never waiting result in products of differing characteristics, then it is use-
for feed, resulting in the best possible overall process ful to know what the product mix will be at capacity pro-
cycle time. duction. If feed is always available to the limiting step,
It is seen from Figure 5 that, for step A cycle time of 1.0 then the fraction of total product produced by a given unit,
h or less, the cycle time of the process is, indeed, the cycle j, in the limiting step is:
time of the limiting step, step B. This step has an equiva-
lent cycle time of 2.4 h. Cmax
However, between a step A cycle time of 13 h, the pro- Fmax, j = (7)
cess cycle time is greater than the limiting-step cycle time cmax, j
(to the left of 2.4 h, step B is limiting, to the right, step A
limits). When the cycle time of step A is 3.0 h or greater, If feed is not always available to the limiting step, then
then the process cycle time is equal to the step A cycle time Eq. 7 is not accurate but can be used as an approximation.
,which is, in fact, now the limiting step.
Figure 5 also shows that, for the case of a process Estimating the capacity of a simple
having steps with units of unequal cycle time, if the next- batch process
to-limiting step is close in equivalent cycle time of the The annual capacity of a batch process comprised of a

74 www.cepmagazine.org August 2002 CEP


series of steps can be estimated from the overall process feed is always available to the step being considered.
cycle time (h/product batch produced), thus: 2. For a step comprised of units of equal cycle times,
the equivalent step cycle time is the unit cycle time divided
8, 760 10 6 ( fs fo w ) by the number of units in the step, given by Eq. 2.
P= (8) 3. For batch processes in which the units comprising a
CP step are of equal cycle times, it appears that matching
The scheduling factor, fs, is a number to accommodate, equivalent cycle times of the steps will result in a process
apart from on-stream time, the mortifying exigencies of with no units with idle time.
plant operations. Further factors can, of course, be included 4. For batch processes in which the units comprising a
such as one for the faction of material that is on-specifica- step are of equal cycle times, the overall process cycle time
tion, among others. appears equal to the limiting-step equivalent cycle time,
If a process consists of a sequence of steps in which and the process capacity can be calculated from the limit-
each step has units of identical cycle times, then CP = ing-step equivalent cycle time using Eq. 8.
Cmax, and Eq. 2 can be used to estimate the limiting equiv- 5. For a step comprised of units of unequal cycle times,
alent-step cycle time. If a process consists of a sequence the equivalent step cycle time is given by Eq. 5.
of steps in which some of the steps have units of unequal 6. For batch processes in which the units comprising a
cycle times, then CP Cmax, and, therefore, Cmax could be step are of unequal cycle times, the overall process cycle
used in place of CP to provide an upper bound for the time may be greater than or equal to the limiting-step
process capacity. equivalent cycle time. This is due to the fact that, under
Take for example, the process described by Figure 4. As certain circumstances, the limiting step may be waiting
noted, the limiting step is step B, which has an equivalent for feed. Therefore, using the limiting-step equivalent
step cycle time of 2.4 h/batch. cycle time in Eq. 8 will provide an upper-bound estimate
Assume the following parameter values: fs = 0.95, fo = of the process capacity. The use of surge capacity can
0.9, w = 500 lb/batch. Then an upper bound for the process help to alleviate this situation.
capacity would be Cmax = 2.4, giving an upper bound of 7. Related to No. 6 above, for a process in which the
1.56 million lb/yr from Eq. 8. If we use the value for the limiting step has units of unequal cycle time, if the next-
process cycle time of 2.67 h/batch determined from the most-limiting step is close in equivalent cycle time to the
True Basic simulation, we calculate a process capacity of limiting step, then reducing the cycle time of the next-lim-
1.40 million lb/yr. iting step could reduce the cycle time of the overall pro-
Therefore, the upper bound, using the simply found cess cycle time.
Cmax, is an overestimate of about 11%. The fraction of total 8. For a limiting step having units of unequal cycle
product produced by the fast unit of step B is estimated times, if this step has waiting time for feed, then in some,
from Eq. 7 as 2.4/4 = 0.60. The accurate estimate of this but not all, cases giving preference to the shorter cycle time
fraction from Figure 4 or the computer program is 0.67, unit of the step will result in a shorter process cycle time,
giving a 10% error. at least for a step of two units

Conclusions True Basic computer program for a


The conclusions that are given below should be viewed two-step process
within the compass of their simplified derivations and are A True Basic computer program was written to simulate
meant to provide a sense of the behavior of a batch pro- a batch process of two steps in sequence in which the first
cess. step is a single unit and second step consists of two units of
1. The equivalent cycle time of a step may be defined as arbitrary cycle times. This program is available on request
the time per batch for the step calculated over a long period from the author via e-mail. CEP

of time and a large number of batches, and assuming that

Literature Cited JAMES L. MANGANARO is a consulting chemical engineer (44 Dodds Lane,
Princeton, NJ 08540; Phone: (609) 924-2750; E-mail: JLJM1@erols.com).
1. Musier, R. F. H., and L. B. Evans, Batch Process Management, His areas of interest are process and product development, mathematical
Chem. Eng. Progress, 86 (6), pp 6677 (June 1990). simulation, and new-opportunity and acquisition evaluation. Manganaros
2. Sharratt, P. N., ed., Handbook of Batch Process Design, Kluwer 30-year background includes process development at the FMC Corp. R&D
Academic (1999). Center, holding an assistant professorship at Manhattan College, and
3. Ku, H.-m., et al., Scheduling in Batch Processes, Chem. Eng. conducting research in General Electrics fuel-cell program. He holds SB
Progress, 83 (8), pp 3545 (Aug. 1987). and SM degrees from M.I.T., and PhD from Rensselear Polytechnic
Institute, all in chemical engineering, and is a P.E. in New Jersey.
A member of AIChE, he has authored numerous papers and patents.

CEP August 2002 www.cepmagazine.org 75

Vous aimerez peut-être aussi