Anda di halaman 1dari 6

Computational model to aid design analytical decisions for the

obtaining of synthetic chemicals


ALFREDO PASCUAL-BERNAD, JUAN MANUEL GARCÍA-CHAMIZO, MARIO NIETO-HIDALGO
Universidad de Alicante
Departamento de Tecnologı́a Informática y Computación
Carretera San Vicente del Raspeig s/n 03690 San Vicente del Raspeig, Alicante
SPAIN
{apascual, juanma, mnieto}@dtic.ua.es http://www.dtic.ua.es

Abstract: The process of obtaining chemicals makes necessary to determine the set of ingredients and synthetic
threads or events that have taken place. Find both ingredients and threads, is the result of a collection of decisions
made and based on the requirements (costs, availability of reagents, process time, energy balances, etc.). From
a conceptual standpoint, it is a design problem which can be solved by analytical scheme of decision making.
Therefore, the particular solution is to establish a solution tree pruning to keep only one way to obtain the sub-
stance. Computational science has provided concepts and methods that support the design decisions of this kind
of processes. In this paper, we propose a computational model for the design assistance based on this model.
The solution is modular and is organized as a tree structure whose nodes are successively simpler sub-problems.
Ordering of the sequence of decisions is the result of setting precedence relationships between the decisions that
correspond to each module (represented by a tree node structure).

Key–Words: Computational chemistry, software architecture, structural tree, essential triplet, decision heuristics,
synthetic chemicals

1 Introduction The problem of making design decisions that re-


quire software systems architecture has undergone
The chemical synthesis is the problem repository of considerable development [1]–[4].
chemical science. These synthesis processes put in
game aspects of the nature of substances, characteri- The practice of computer science uses the sci-
zation of potential ingredients and phenomenology of entific progress that has provided over the past cen-
the reactions which have to take place. So, the cen- tury meta-science (axiomatization of theories, model-
tral problem of chemical engineering is to obtain sub- ing paradigm, corroborating legitimacy, etc..) as tools
stances which have industrial interest; as well as as- for inspiration for solutions to design problems and
pects indicated in the preceding paragraph, the prob- mills systems [5], [6].
lem incorporates aspects like profitability, sustainabil- From a computational approach, we consider the
ity and, in general, the whole context of occurrence of problem of designing methods to obtain chemicals as
processes. a case of the activity of ”know”. It is the notion that
This framework of issues, circumstances, actions you get to the question ”why” we want to serve some-
and events is complex. Far from the optimal solution, thing, or why we want its knowledge. And it refers to
what can hope for is to find good enough solutions, i.e. the performance or functionality that has that knowl-
optimized. Being of type NP-complete, the problem edge, in this case the synthesis of the substance.
of decision-making underlying the chemical synthesis Our problem has a nature of process or method,
has a solution that is beyond the capability expected and we are interested above all ”know how” because
systematization of mathematics. So, is appropriate of it specifies the ”know”. This is made with the re-
to resort to the incorporation of heuristics specific to striction of knowledge when it contains the constitu-
computer science. tive aspect leading to the service or function, i.e. rel-
We seek to develop software architectures with ative to the structure and organization. It combines in
support for decision making, to design processes to itself two aspects of ”know”: notion and operational.
obtain chemical test substances. We can define the By ”know how”, i.e. knowledge about the facts or
chemical processes as consisting of activities, prop- things created with this increased role of the human
erly organized, which form chemical phenomena. intellect, is what we call ”knowledge”.
2 Classification of Design Factors of
a Chemical Process
The specification of the test substance contains, ex-
plicitly or implicitly, the requirements to be met by the
chemical process of obtaining it. Shares of chemical
process design must aim to meet the specified require-
ments, together with all other factors to be involved in
the chemical process [12], [13].
Indeed, the statement ”come from the require-
ments of the problem” provides a first classification
of the factors involved in chemical process design, so
that you get two kinds of factors: essential and cir-
cumstantial.
The essential factors of the chemical process de-
sign part of the intimate nature of it (the performance
of the compound, its structure, etc..), Ie the factors
constituting the result. For example, are essential fac-
tors in the development of ”water substance” hydro-
gen atoms, the reaction pressure, the ratio of the mix-
Figure 1: Schematic design of a chemical process ture of ingredients, the heat put into play; among other
Situational factors are who make the context of
chemical process design (cost, author, time plan, lo-
It still remains the action ”know how to do,” that cation, etc..).
we might express as ”know how”. The meaning is In general, the set of decision factors is:
to produce the synthesis of the chemical. It includes,
therefore, the activity of conceiving the ingredients θ = {pressure, temperature, proportion,
which chemical reactions will lead to solving the
concentration, velocity, catalysis,
problem. To the questions ”why” and ”how” above,
it adds the question ”how” whose responses give no- ingredients, sub − products, targetcompound,
tice of the ingredients, the constitution and usefulness price, placeof production...} (1)
of the facts or things.
There is a requirement which suggests that the so- The essential factors, clearly, are:
lution is constructive, as directly relevant to the design
of chemical process, and equivalent to the question ”to ε = {pressure, temperature, proportion,
do”, answering ”to know how to do.” Reply to ”to do” concentration, velocity, catalysis,
is to make design decisions through an analytical pro- ingredients, sub − products, targetcompound} (2)
cess. Once the design of a chemical process analyt-
ically, its implementation may consist in a synthetic Situational factors are:
process that is contained in the ”know how to do” re-
sulting from the design itself. Γ = {price, placeof production} (3)
In addition to questions that lead to the concep-
tion of the process, it corresponds to answer questions Formally, the sentence ”come from the require-
related to the context of the ”when” (produce method- ments of the problem” defines an equivalence relation
ological work plans, time, etc..), ”who” will do (which among the factors making the chemical process de-
give us the criteria for defining the work teams), ”how sign and, therefore, provides a partition between these
much” (clearly concerned with the budgetary aspects), factors.
and many others. We assume that the set of design factors of the
Figure 1 shows schematically the process of de- process is:
signing a chemical process: starting from a concept,
the task is to conceive the chemical process in all Φ ≡ {ϕ1 , ϕ2 , ϕ3 , ...ϕi , ...ϕn } (4)
its aspects (design) to get a guide (implementation
project) to the test compound (realization). Many are The linkage of each design factor is the chemical
the proposals to provide environments to aid design process or a part thereof, or in which context is a fac-
decisions [7]–[11]. tor. If the engineer found that one factor is likely to
be considered equally essential as circumstantial fac- factors, those that express the functionality that ex-
tor will have to decide the allocation of a value (by hibit chemical processes through its interface with the
definition). This purity that may seem inconvenient in world.
practice provides a powerful heuristic decision. Let The sentence ”degree of substantiality” of a fac-
us assume that the price of a compound is a circum- tor in relation to the aforementioned process defines
stantial factor from the usual approach, however, if the equivalence relation between the essential factors.
the feasibility of obtaining the compound on the price, Similarly as stated in the case of the factors creating
there will clearly be regarded as an essential and only processes in general, if an essential factor to be con-
essential factor. sidered equally supported as belonging to more than
Having made these considerations, the demon- one class, will be decided by definition, assigning
stration that the sentence ”come from the require- it to one of them. For example, a genuinely intrin-
ments of the problem” establishes an equivalence re- sic factor, such as the ingredients of a reaction, you
lation among the factors making the chemical process agree to be defined as explicit factor if these consti-
design is sufficiently immediate and therefore we have tute a pre-determined design of chemical processes
ignored in this document. [14], [15]. This would apply to design the process
So, we have: to absorb carbon dioxide using lime, lime, an ingre-
dient that could be described as intrinsic factor, will
θ =ε∪Γ (5) be classified as an explicit factor in this process be-
cause the requirements explicitly state that part lime
the process. Therefore, include lime, far from being
ε ∩ Γ = {} (6) decidable, introduces a direct pruning the tree of pos-
sible processes that absorb carbon dioxide.
All the key factors are:
Φ/R = {ε, Γ} (7)
ε ≡ {1 , 2 , 3 , ...i , ...m }, , (8)
We formally justified application of the tech- i = ϕi , m ≤ n
nique ”divide et vinces” to the problem of making
design decisions in a chemical process, obtaining The demonstration that the ratio is equivalence,
two sub-problems of smaller size: the essential sub- it’s so simple enough that has not been included.
problem on essential design factors, and the context The consequence is that all of the essential fac-
sub-problems, which refers to the situational factors tors to be partitioned to support the relevance criteria
that shape the environment in which chemical process to be accorded each factor. The definition of these
takes place. criteria of relevance is a systematic tool making de-
sign decisions. The degree of substantiality can be set
according to different criteria and different classifica-
3 The Essential Factors tions result.It is especially interesting considering the
The essential factors, as we have said incorporating particular case of the three ranks of substantiality we
aspects ”technology”, ”organization” and ”benefits” mentioned in the preceding paragraphs because that
that are recognized in the processes. The answer to classification of the essential difference in the factors
”how” to give effect to a process determined technol- of the architecture of the structure (since both of the
ogy to use, i.e., the components of the process: ac- technology).
tions. The answer to ”how to” encompasses the pro-
cesses structure and the organization that are to com-
pose actions. Reply to ”why” we want the process es- ε = A ∪ E ∪ T, , (9)
tablishes the aspects of performance. Obviously, the
latter is the overall level resolutions and decisions, as A ∩ E = A ∩ T = E ∩ T = {} (10)
such, is essentially to established requirements.

ε/GS = {A, E, T } (11)


3.1 Classification of the Essential Factors
The essential factors of chemical processes can be de- The essential design sub-problem admits, there-
scribed as intrinsic factors, i.e. stocks and the entities fore, be broken down into three subproblems, each of
over which actions occur. Or may be factors involved which specializes in the areas of architecture (includes
on the structure of processes and guidelines concern- explicit factors can be identified by asking ”why”),
ing relationship between actions. Or are the explicit technology (the intrinsic factors obtained by asking
”how”) and structure (the factors involved, the ques- 4 Generalization of the Structural
tion ”how”).
Tree
The objective of this paper is applied to provide the
A = {targetcompound, sub − products...} (12) conceptual basis for the development of a support-
ive environment for decision making modular design
[16]–[18].
E = {proportion, velocity, The tree structure and sequence of actions estab-
concentration, velocity, catalysis...} (13) lished by the triplet are the essential core of a recur-
sive strategy to support that environment. Just con-
sider the structural tree leaves obtained to a step in
T = {ingredients, pressure, temperature...} (14) the design process of the chemical process, which are
sub-problems, in turn, as problems in themselves and
Therefore, the essential factors of production of solve them according to their own essential triplet.
a device may be obtained by what might be called Recursively, the process continues until leaves whose
”triplet essential”. solution is known. The sequence of design decisions
due to the pattern of causation or precedence of atten-
tion that corresponds to the kinds of design factors of
T ripletEssential = hA, E, T i (15) the chemical process.
The solution obtained by this procedure, as it ig-
3.2 Management of the Essential Factors nores the specifics of the problem starting, leaves you
far from optimal. Therefore, in the subsections that
It is pertinent to question the desirability of finding
follow, we propose an alternative strategy for that spe-
criteria for the proper sequence to proceed. To this
cific aspects of the problem will be of assistance in
end, the ”causality” can establish a relationship of or-
making design decisions.
der, whose simple proof was not transcribed in all the
classes derived from the degree of substantiality are
essential factors in knowledge. 4.1 Problem Dependent Decision Process
A key factor is the cause of (or precedes) another
one if it participates in the construction of the second. In this section we add the tree level structural spec-
Every factor is the cause of itself. ification based on criteria that are based on specific
aspects of the chemical process we are interested in
design. For each node of the last level is required to
∀i , j ∈ ε, , ([(i )]) ⇒ j ) ⇒ i ≺ j (16) propose an equivalence relation which allows further
divided into nine sub-problems. Iteratively, the pro-
The semantics of causality produces ultimately an cess continues until the sub-problems represented by
expression of subordination between the three kinds the tree’s leaves are known solution.
of essential factors: On the set of factors may define a node poten-
tially more of an equivalence relation and will need to
choose one of them to continue building the tree struc-
T ≺E≺A (17) ture. This choice will influence what the solution is fi-
nally. This circumstance provides heuristic optimiza-
The sequence features, structure and technology tion design proposal, whose essence is to evaluate the
is to establish the performance that has the talent characteristics of various structural sub-trees resulting
to meet with the task of achieving this functionality, resolution whichever is the equivalence relation that is
decide on the appropriate use and organization ulti- used at each level and the sequence actions.
mately wants to know what it is and how elaborate In the absence of equivalence relations between
ingenuity decide what are the right ingredients to get all of the essential factors of the sub-problem corre-
it. It is the action sequence that corresponds to the an- sponding to a given node, you can use other criteria
alytical methodology for the design of chemical pro- (pseudo-relations) approximations leading to qualify-
cesses, which should be ”top down”. The analytical ing not disjoint. The resulting algebraic structure is a
method is causally consistent. graph.
As for the sub-problems that arise in applying
each new equivalence relation, spatial corresponding
A → E → T, , AET (18) set of classes will determine the sequence of actions
to follow in the decision making process to propose
the solution to the problem.

4.2 The Design Context


The identification of situational factors may be re-
sponding to questions about context implicit in the
”to do”, that is, remaining after excluding those of the
triplet essential. By analogy, we call ”contextual tu-
ple” a collection of questions to understand.
hwhy, when, who, howmuch, wherei
The tuple context potentially can cover a wide
area and include factors unique to specific questions
or problems. In the general case is considered ade-
quate contextualization comprising a compendium of
phrases descriptive of the motivation behind the cre-
ation of knowledge, the time frame envisaged for the
plan to propose the solution process and to validate the
result, the description of the team the budget, the real-
ization of the equipment and facilities, and indicators
of the progress of creative actions.
Unlike the case of the essential factors, it is easy Figure 2: Tree structural design of the production sys-
to understand that it is pointless to propose a general tem of a chemical plant. Solution based on the equip-
criterion qualifying questions in the context tuple by ment required.
its very nature.
For the sake of this systematic care while method-
ological consistency, the option is to undertake the
sub-problem context of solving a problem and, in turn,
a problem. From there, proceed recursively. Situa-
tional factors that are most relevant for solving each
particular problem will be to acquire the quality of es-
sential factors of the problem explicit contextual, and
so on.

5 Environment to Support Design


Decisions
Since the proposal of the preceding paragraphs we
have developed a prototype computer tool to aid de-
sign decisions chemical processes. From the require-
ments in the process specification problem, we define
the variables that characterize them and their value
ranges. For each of the variables define an equiva-
lence relation and we obtain the quotient set.

5.1 Case Study


The case that we propose corresponds to a production Figure 3: Tree structural design of the production sys-
system chemical plant, which consists of a chemical tem of a chemical plant. Solution based on the mix-
reactor in which equipment asset depends on the func-
ture.
tionality that you want to assign.
By example, if the reaction mixture is homoge-
neous and flow works continuous, then it is only nec-
essary to activate the chemical reactor. However, if the
reaction mixture is heterogeneous or if one of the em- Networks and Distributed Systems. 6(2), 2011,
ployment opportunities is one contemplates discontin- pp. 142–157
uous load reactor, then it is necessary to activate other [6] S. C. Kleene, Introducción a la meta-
instances of equipment to ensure the viability of the matemática. 1974, Ed. Tecnos
process. Figures 2 and 3 show two possible struc- [7] L. Lee and P. Kruchten, A Tool to Visual-
tural tree to solve the problem. ize Architectural Design Decisions, Proceedings
of the 4th Int. Conf. of Quality of Software-
Architectures: Models and Architectures. 2008,
6 Conclusion pp. 359–362
We were able to reach a solution to the problem, al- [8] L. Lee and P. Kruchten, Visualizing Software
ways from a functional specification of the require- Architectural Design Decisions, Proceedings of
ments must meet. Clearly defined key factors and also the 2nd European Conf. on Software Architec-
circumstantial, we could also justify making the se- ture. 2008, pp. 43-54
quence decisions related to these two classes of fac- [9] A. Tang, Y. Jin and J. Han, A Rationale-based
tors surrounding our problem. Architecture Model for Design Traceability and
The next step is obvious: the generalization of Reasoning, J. of Systems and Software. 80(6),
the procedure, so that we provide solvent technique 2006, pp. 918–934.
in making analytical design decisions to obtain syn- [10] A. Tang, J. Han and R. Vasa, Software Ar-
thetic products chemicals. Applied successively to chitecture Design Reasoning: A Case for Im-
parts of the problem of such a design, and the sub- proves Methodology Support, IEEE Software.
problems are obtained that conform to achieving con- 26(2), 2009, pp. 43–49
trol of the issue. Since the implementation of this pro- [11] J. S. Van der Ven et al., Design Decisions: The
totype computing environment decision support has Bridge between Rationale and Architecture. Ra-
been demonstrated by a case potential of study. tionale Management in Software Engineering.
The solution provided the design of working con- 2006, Springer.
ditions in a chemical synthesis, without having to be [12] M. Che and D. E. Perry, Scenario-Based Archi-
the optimal, it leaves open the door to work to opti- tectural Design Decisions Documentation and
mize the solution. It resorts to the need to establish Evolution, IEEE Software: 18th IEEE Int. Conf.
criteria justified, and propose an objective function. and Workshops on Engineering of Computer-
Finally, the formal framework presented in this Based Systems. 2011
paper is the core of a broader approach that aims to [13] M. Shanin, P. Liang and M. R. Khayyambashi,
provide conceptual support for the creation of scien- A Survey of Architectural Design Decision Mod-
tific knowledge and linking bodies of knowledge of els and Tools. 2009
ICT to those who make up the Chemistry. [14] C. W. Gardiner, Handbook of Stochastic Meth-
ods: For Physics, Chemistry and the Natural
Sciences. 2002, Springer Series in Synergetics.
References: [15] P. Van der Ween, J. M. Baetens and B.
[1] P. Kruchten, R. Capilla and J. C. Dueas, De Baets, Design and Parameterization of a
The decision views role in software architecture Stochastic Cellular Automaton Describing a
practice, IEEE Software. 26, 2009, pp. 36–42 Chemical Reaction, J. of Computational Chem-
[2] N. Medvidovic and R. Taylor, Software archi- istry. 32, 2011, pp. 1952–1961
tecture: foundations, theory, and practice. Pro- [16] R. Capilla et al., A Web-based Tool for Manag-
ceedings of the 32nd ACM/IEEE Int. Conf. on ing Architectural Design Decisions, ACM SIG-
Software Engineering ICSE ’10. 2010 SOFT Software Engineering Notes. 31(5), 2006,
pp. 4–11
[3] R. N. Taylor, N. Medvidovic and E. Dashofy,
[17] A. Jansen et al., Tool Support for Architectural
Software Architecture: Foundations, Theory,
Decisions, Proceedings of the 6th IEEE/IFIP
and Practice. 2009, John Wiley and Sons. New
Working Conf. on Software Architecture. 2007,
York.
pp. 44–53
[4] D. Tofan, Tacit architectural knowledge, Pro- [18] F. Naval, R. Capilla and J. C. Dueas, Processes
ceedings of the Fourth European Conf. on Soft- for Creating and Exploiting Architectural De-
ware Architecture ECSA 2010, pp. 9–11 sign Decisions with Tool Support, Proceedings
[5] M. Dbouk, A. Sbeity and H. Mcheick, Towards of the 1st European Conf. on Software Architec-
service-based approach: building huge software ture. 2007, pp. 321–324
architectural design, Int. J. of Communication

Anda mungkin juga menyukai