Professional Documents
Culture Documents
Evaluating the Complexity of Petri Nets and Ladder Logic Diagrams for Sequence
Controllers Design in Flexible Automation
Abstract - The varying control specifications of developed a commercial product based on an enhanced PN
today's flexible and agile manufacturing systems are that interacts with the physical system to control [8].
challenging the use of ladder logic diagrams (LLDs) and More details and advantages of PNs for controlling
high-level programming language methods. More manufacturing systems can be found in [1,11.17.20].
recently, Petri nets (PNs) are increasingly used to design More recent studies of PNs for discrete event control can
sequence controllers. However, to establish PNs as an be seen in 16.12-20).
alternative to LLDs there is a need to justify the PN The detailed comparison of LLDs and PNs is very
method. This can be partially accomplished by comparing important in realizing their advantages and disadvantages
the complexity of PN and LLD designs. This paper and, particularly. in establishing PNs as an emerging
presents a methodology to evaluate the complexity of PNs design technique for effective sequence control of
and LLDs for sequence controllers design. Complexity of industrial automated systems. Realizing this, we
a design is characterized by the number of basic elements identified certain criteria to compare LLDs and PNs
used to model the given control hgic. More specifically, through a particular industrial automated system [16].
this paper presents analytical formulas to estimate the Based on the number of basic elements used in designing
number of basic elements to model certain building blocks sequence controllers. it was claimed that PNs offered a
of logic modeling when PN and LLD w e used. Also, by better solution than LLDs. It is reported that the fewer the
presenting a methodology to use these analytical basic elements in a controller, the better the model used to
formulas, this paper precludes the need for physically design the controller [1.9.10,16]. The present work is
building the controllers by either PN or LLD for the continuation of the earlier work reported in [16].
comparison of their graphical complexity. The results are Motivated by the fact that counting the number of basic
demonstrated by considering two examples of sequence elements in a sequence controller becomes cumbersome
controllers. Finally. the limitations of the present work when the control specification becomes complex, this
are presented along with possible extensionv for future paper presents analytical formulas that assist to estimate
research. the basic elements in a controller.
The goal of this paper is to formalize the comparison
INTRODUCTION between PNs and LLDs by presenting a methodology to
compare the graphical complexity of PN and LLD without
Ladder logic diagrams (LLDs) are widely used to actually modeling the sequence controllers. The
design sequence controllers in manufacturing systems objectives of this paper are:
[6.7]. The sequence controller in this paper means a class 1. To formulate certain mathematical formulas to
of discrete event controllers without choices in executing calculate the number of basic elements to model
the operationslactivities. LLDs specify the U 0 procedures certain building blocks of logic models using PNs and
of the Programmable Logic Controller (PLC) that drive and LLDs,and
perform the repetitive operations of the system. These 2 . To present a methodology that synthesizes these
diagrams grow so complex that locating the cause when a analytical formulas for estimating the total number of
problem is detected becomes extremely difficult [ 1.2.1 61. basic elements needed to design sequence controllers
Furthermore their usage is limited only to control the using PNs and LLDs.
system but not to analyze and evaluate the qualitative and
performance characteristics. Also, they often need LOGIC DESIGN BY LLDs AND PNs AND THEIR
significant changes as the specification changes. Due to COMPARISON CRITERIA
these reasons modem discrete manufacturing systems have
challenged the use of LLDs to design sequence controllers. The application of LLDs for sequence control is
Hence, researchers are constantly pursuing IO develop widely known as they are used by several industries [3,10].
integrated tools that eliminate the limitations of LLDs. An excellent tutorial on PNs and their applications is
These tools are aimed not only for control but also system given in 171 and their applications in manufacturing
analysis, evaluation, and simulation. Petri nets (PNs) automation are reported in [4.11,20]. The systematic
originated from the field of computer science are such a methods to formulate PN models can be seen in [18-20]
tool as evident from its versatile applications in and the methods of developing LLDs can be seen in [lo].
manufacturing systems [ 11,191. Realizing the potential By incorporating input/output and timing information to
of PNs for control, in France, GRAFCET - a PN like places and transitions, Real-time PNs (RTPNs) were
representation tool is proposed as a standard specification introduced in 116). Physically, RTPNs are same as PNs
of sequence controllers [3]. Its international standard except that in the former. places are given attributes to
version is called sequential function charts 151. Hitachi model input sensory information and transitions are
429
indicates that PN is always preferred to model timed the earlier section. Hence. in order to find the basic
logical AND. elements to model a given control logic using the models
developed, first it has to be decomposed into the logic
Timed Logical OR constructs or segments. Then for each segment, the
corresponding analytical formula is applied to find the
basic elements. The basic elements corresponding to all
In this case: a = 3d + n + dn. p = 8d + 3n, andd =
such segments are then added to obtain the total number of
fld,n) = -5d - 2n + dn. This indicates that for d and n values
which satisfy 5d + 2n > dn. PN is preferred. For d and n basic elements (p) needed to model the given control logic
values that do not satisfy the above condition, LLD is using a LLD.
i=k
preferred. For example, for the pairs, d = 3 . n = 4 ; and d =
4 , n = 3, PN is preferred. On the other hand for d = 7. ti = P= E , Pi
i=l
8, LLD is preferred. Similar analysis can be performed for where P =Total number of basic elements to model the
any given d and n values using A function. control logic using PN,
Pi = Number of basic elements in segment i, and
Timed Sequential Modeling
k = Total number of segments.
The final step (as shown in Fig. 2) is slightly
In this case: a = 4d + 2 (m + n ) - 3 , p = 1Od + 2m + 3n
different in the case of PN due to its physical model. This
- 5. and A = fld,m,n) = -6d - n + 2. A indicates that PN is is because when modeling a control logic, intersegment
always better than LLD. conditions and connections exist in a PN model. They are
explained below.
METHOD T O USE THE ANLYTICAL FORMULAS For some cases in the given control logic, the output
condition for segment i may be one of the input
Figure 2 illustrates the method to obtain the total conditions for segment j . These conditions are called as
number of basic elements to model a given control logic intersegment conditions. In LLD. such conditions are
using the analytical formulas. This is described below. separately modeled in both segments i and j . Even though
such conditions appear in more than one segment, the
place modeling such condition physically appears only
once in the PN. Therefore. in case of PN to account for the
repetitive count of these intersegment conditions, the
total number of such conditions has to be subtracted from
Divide it into logical constructs, logical AND, the summation of basic elements in all segments.
logical OK, sequential model, etc. Also, for some control specifications it may possible
that an output condition of segment i causes an action
which produces an input condition(s) for segment j . That
431
Sometimes, i t may happen that an action assembly station. After the assembly is completed. the
corresponding to the output condition in segment i robot unloads the finished product on a work table. The
produces several inputs for segment J. Considering this, AGV transports empty pallets from pallet storage station
e;,] is given as follows: to assembly station. Robot is also used to unload the
e$ = q + zw,.
r=l
finished product from work table and place it on the pallet
which is on the AGV. Finally, the AGV transports the
finished product loaded on pallet to inspection station for
where w,.represents the number of inputs for segment j inspec tion.
produced by output condition r in the segment i. The first
m u
term in 8 . . corresponds to the output arcs from the places lllU storag
1Jl utioo
modeling output conditions in segment i and the second
term corresponds to the input arcs to the places modeling
input conditions in segment j . Hence, 01J. . is given as:
switchD L,,,it Suaor C
0-=2q+?w,.
,.=I
The total number of basic elements modeling intersegment
switch E
n
connections in the whale control logic is then given as
Robot
follows: switch F
i=$= keij
e=
i#J
A B D
- x +
6xl
+I+
I+ I
Timer
2m+2n+ 2m+3n =
2m+2n+ 2m+3n =
2m+2n+ 2m+3n =
1=11
Figure. 4(a) PN model of the system. (b) LLD of the system m=2,n=3
2m+2n+ 2m+3n =
Now, consider another system where intersegment 1= 9
connections exist in the control specification. This is m=? n = l
illustrated in the next example.
Here, @ = 0, since there are no intersegment
Example 2
conditions. fl is calculated as follows:
The system considered consists of four pneumatic Upon observation it can be seen that intersegment
pistons (A, B, C, and D) which are to be sequenced by connections exist between segments 1 and 2; 2 and 3; 3
double activated five ports and two-way solenoid valves. and 4; and 4 and 1. For example, A+ and B+ in segment 1
results in a+ and b+ in segment 2. Therefore for segments
Each piston has two normally open limit switches. In
1 and 2, q = 2. Assuming A+ is action 1 and B+ as second
flexible automation, typical functions of these pistons
can be to load/unload parts from a machine table, to 6 1 2 = 2q + W I + w~ = 2(2) + 1 + I = 6. Similarly, for
extend/retract a cutting tool spindle, to feed a bar-stock, segments 2 and 3, 823 = 6. For segments 3 and 4
etc. A typical pneumatic piston X (can be A. B, C. and D) assuming B-, D-, C+ as first, second, and third actions,
is shown in Fig. 5 . When the end of piston X contacts
limit switch x l (x0). x l (x0) is closed indicating that the 634 = 2q + W I + w2 + w3 = 2(3) + 1 + 1 + 1 = 9. For
piston X is at the end of its forward stroke (retum stoke). segmenu 4 and 1. 641 = 2q + WI = 2(1) + 1 = 3. Hencc. 6
A push button. START is provided to start the system. It = eI2 %3 + e,
+ +e4, = 6 + 6 +9 + 3 =24.
is required that the pistons A. B. C. and D have to be
433
Hence, a $, and A are given as follows: This indicates for the example system considered.
a = (9 + 9 + 11 + 9) + 24 = 62, LLD gives 32.25 3% shorter program than PN. This can
also be validated by manually counting the basic elements
p = 10 + 10 + 13 + 9 = 42. and after formulating the physical models of PN and LLD. The
A=a-p=20. PN and LLD for to control the system is given Fig. 6(a)
and 6@) respectively.
CONCLUSIONS
434
of basic elements to model the most commonly used Maintainable Sequence Control and its Applications
building blocks of logic modeling by both PN and LLD. in Factory Automation." IEEE Trans. on Industrial
Furthermore, a methodology that uses the developed Electronics. 33 (1). 1-8. 1986.
analytical formulas to estimate the total number of basic [91 Pessen. D.W.. The Design and Application of
elements to model a control logic is presented. The Programmable Sequence Controllers for AIctomation
concepts developed in this paper are demonstrated by Systems, Longman Inc., New York, 1979.
considering two examples of sequence controllers. The [lo] Pessen, D.W.. "Ladder-Diagram Design for
examples considered here demonstrates the practical Programmable Controllers." Automutica, 25 (3).
application of the research results. The methodology 407-412. 1989.
provides an accurate quantitative comparison of PN and [Ill Silva, M.. and Valette. R.. "Petri nets and Flexible
LLD in terms of basic elements. By precluding the need Manufacturing." Advances in Petri nets , Lecture
for physically building the controllers by either PN or Notes in Computer Science, Springer-Verlag.
LLD. this methodology serves as an effective aid for a Heidelberg. 374417. 1990.
control engineer to select between PN and LLD even before 1121 Stefano, Di. A., and Mirabella, 0.."A Fast Sequence
starting to write the control program. However, in case of Control Device Based on Enhanced Petri nets,"
complex control specifications, decomposing the control Microprocessors and Microsystems, 15 (4), 179-
specification in terms of the logic constructs may be 186. 1991.
difficult. However. this problem can be solved using the [I31 Valette, R.. Courvoisier. M.. Bigou. JM., and
methods such as Kamaugh maps. Huffman method. etc. Albukerque. J., "A Petri net Based Programmable
The work related to estimating the number of basic Logic Controller," Computer Applications in
elements in specifications that include the Production and Engineering. EA. Warman (editor),
implementation of emergency stop, counters, and relays is North-Holland Publishing Company, IFIP, 103-115.
in progress. 1983.
[14] Venkatesh. K., Subramanian, C., and Masory. 0.. "A
REFERENCES Sequence Controller Based on Augmented Timed
Petri Nets," Proc. of the 6th Annual C o d . on Recent
Boucher. T.O.. Jafari. M.A.. and Meredith, G.A.. Advances in Robotics, Gainsville, FL. 1-6, 1993.
"Petri Net Control of an Automated Manufacturing [ 151 Venkatesh, K., Kaighobadi Mehdi. MengChu Zhou,
Cell." Computers in Industrial Eng., 17 (1-4). 459- and Reggie Caudill. "Augmented Timed Petri Nets for
463, 1989. Modeling of Robotic Systems with Breakdowns,"
Chaar. J.K., Voltz, R.A., and Davidson, E.S.. "An Journal of Manufacturing Systems. Vol. 13, No. 4,
Integrated Approach to Developing Manufacturing pp. 289-301. 1994.
Control Software". Proceedings of the 1991 IEEE [16] Venkatesh. K.. Zhou. M.C.. and Caudill, R.,
I n t . Conf. on Robotics and Automation. "Comparing Ladder Logic Diagrams and Petri Nets
Sacramento, CA, 1991. for Sequence Controller Design through a Discrete
David, R., From Grafcet to Petri Nets , Englewood Manufacturing System,'' to appear in IEEE Trans. on
Cliffs, NJ: Prentice Hall, 1992. Industrial Electronics, 1994.
DiCesare F.D.. and Desrochers. A.A.. "Modeling, [I71 Venkatesh. K.. Zhou, M.C., Caudill, R., "A Control
Control, and Performance Analysis of Automated Software Design .Methodology for CIM Systems,"
Manufacturing Systems using Petri nets," Control Proc. of Rutgers Conf. on CIM in the Process
and Dynamic System, Vol. 47. C.T. Leondes (Ed.), Industries, East Brunswick, NJ. April 25-26, 565-
New York Academic, 121.172. 1991. 579, 1994.
Falcione. A., and Krogh, B.H.. "Design Recovery [I81 Zhou, M. C., F. DiCesare. and D. Rudolph, "Design
for Relay Ladder Logic," IEEE Control Systems and Implementation of a Petri Net Based Supervisor
Magazine, April, 90-98, 1993. for a Flexible Manufacturing System," IFAC Journal
Ferrarini, L., Narduzzi, M., and Tassan-Solet. M.. "A Automutica, 28 (6). 1199-1208. 1992a.
New Approach to Modular Liveness Analysis [I91 Zhou, M.C., DiCesare, F.. and Desrochers. "A
Conceived for Large Logic Controllers' Design," Hybrid Methodology for Synthesis of Petri net
IEEE Trans. on Robotics and Automation, 10 ( 2 ) . Models for Manufacturing Systems." IEEE Tram. on
169-184. 1994. Robotics and Automation, 8 (3). 350-361. 1992b.
Murata. T.. "Petri Nets: Properties, Applications and [20] Zhou, M. C.. and DiCesare, F.. Petri Net Synthesis
analysis." Proc. of IEEE , 77 (4). 541-580, 1989. for Discrete Event Control of Manufacturing
Murata, T., Komoda. N.. Matsumoto. K., and Haruna. Systems, Kluwer Academic Publishers, Boston, MA,
K., "A Petri net-based Controller for Flexible and 1993.
435