Creative Problem Solving in ED
Roni Horowitz
Roni Horowitz (1999), Creative Problem Solving in Engineering Design, (Doctoral dissertation, Tel-Aviv Univ.).
Chapter 3. Sufficient Conditions for Inventive Designs
Problems arise at different stages of a product’s life cycle. It is generally accepted that creativity is mostly required in the initial stages [Ulrich, 88, Wiliams, 90, Gero, 89] where the number of constraints is still relatively low and many alternative design concepts can be explored. But creativity is also needed to solve problems that crop up in more advanced stages such as detailed design, process planning, production or service where an engineering system already exists. In these stages conceptual changes, though sometimes unavoidable, are generally not desirable. This research deals with such problems. Altshuller classifies the solutions to these problems as “inventions within a paradigm” [Altshuller, 1985], while Arciszewski calls them ‘improved solutions’ [Arciszewski, 1995]. ARIZ (TRIZ’s problem solving algorithm) makes a distinction between the “mini-problem” in which one tries to solve a problem while keeping to a minimum the modifications made to the existing system, and the “maxi-problem” in which large (often conceptual) modifications are sought. The focus of this dissertation is thus the mini-problem. (pp. 42-43)
3.1.1. Altshuller's Theory
One of the few attempts to characterize design inventions in terms of objective necessary and/or sufficient conditions was carried out by Altshuller [Altshuller, 1985]. Altshuller conducted a comprehensive study of a large body of data stored in patent collections. His main finding was that a necessary condition for design inventions is that they incorporate an ‘elimination of a conflict’. Conflicts in engineering systems arise between a system parameter that should be improved to meet the requirements, and another system parameter which inadmissibly deteriorates as a result of the improvements. Consider for example the conflict in the design of an incandescent light bulb: on the one hand a requirement for efficient energy consumption dictates high filament temperatures, while on the other hand filament temperature should be kept low to ensure the bulb’s long life.
Routine engineering deals with a state of conflict through trade-off - a search for the best compromise between the conflicting requirements. Designers of filaments, for example, try to identify the filament temperature that would best reflect customers’ requirements. Elimination of a conflict means resolving the conflict under the condition that a compromise is unacceptable. For the filament conflict, this would mean improving both energy consumption and durability. Because Altshuller’s criterion of conflict elimination is one of very few attempts for objectively characterizing inventions* in terms of their internal features (patent offices, for example, use an ‘external’ criterion based on the relation between the examined solution and other known solutions), and because it paved the way for the development of the successful TRIZ method [Fey, 1994; Sushkov 1995; Lirov, 1990] for supporting creative design, we used the idea of ‘elimination of conflict’ as a starting point for our research.
*Another criterion based on the concept of logical distance was suggested by Arciszewski et al. (p. 45)
3.3.2. The Qualitative Change (QC) condition
Any solution in which at least one problem characteristic changes from an increasing relation to either a decreasing or a neutral relation is said to incorporate a qualitative change The Qualitative Change condition is formally presented in Expression (1) in which r is defined as a pair of undesired effect (UDE) and a related attribute; P is defined as the set of all problem characteristics.
(1) ∃ r ∣ r∈Pproblem ∧ r ∉Psolution
Expression (1) reads as follows: there exists in the problem state a pair, r, which is a problem characteristic and that r is not anymore a problem characteristic in the solution state. A UDE-attribute pair can cease to be a problem characteristic only if the undesired effect becomes totally insensitive to the value of the attribute (with which it shared a problem characteristic) or if the intensity of the undesired effect becomes a decreasing function of that attribute). (p. 53)
3.5. The rationale behind the sufficient conditions
We show, by resorting to other theories of creativity and engineering design, that the sufficient conditions are in line with other criteria and conditions for originality and usefulness of a solution to an engineering problem. As mentioned in chapter 1 the combination of usefulness and originality is considered by most of creativity researchers as a necessary and sufficient condition for a creative solution.
3.5.1. The Relation Between The Conditions and The Originality of the Solution
3.5.1.1. Functional Fixedness
Functional Fixedness, which was described by Duncker [Duncker, 1945], has since been subject to intensive research in cognitive psychology. Functional Fixedness is a state in which the problem solver cannot conceive of uses for an object apart from its normal use. Ulrich [Ulrich, 1988] mentions Function Sharing, a design process in which an object already carrying out one function is assigned another one, as one of the processes that constitute inventive design. But to identify new functions to be assigned to an object one has to over-come functional fixedness, and few do. The CW condition, by not allowing the introduction of new objects into a system, forces the problem solver to achieve the new functionality needed for a solution using existing objects only. The problem solver is thus forced to search for a solution in areas that others would probably overlook due to functional fixedness. Thus solutions that comply with the CW condition, being rarely attained, are by definition original. (pp. 64-65)
3.7. Summary of Chapter 3
An engineering solution satisfying the conditions manifests a delicate balance in its distance (in terms of the amount of modifications made on the structure of the engineering system) from the problem situation. The Closed World condition dictates a minimum number of modifications whereas the Qualitative Change condition requires that the solution exhibit a qualitatively different behavior which naturally calls for large structural modifications. The conditions determine the boundaries of a narrow gap in the solution space, overlooked by most problem solvers, in which (almost) the same structure exhibits a qualitatively different behavior. (p. 75)
4.1. The underlying principles of the SIT method
As mentioned above the SIT method comprises five idea provoking operators (which will be explained below). In order to apply the operators the problem solver must prepare a problem formulation that includes the QC and CW conditions. Two main stages thus make up the SIT method: preparation : problem formulation and analysis and solution: problem solving. The preparation stage comprises three sub-stages:
A. The problem’s closed world is defined by identifying the types of objects that form the given engineering system and those residing in its neighborhood.
B. To have a better understanding of the rational underlying the system’s design, the problem solver is guided to construct a hierarchical functional model of the system (explained below). The functional model is required only when using the Object Removal technique (when an object is removed from the system its important to determine its exact function in the system) and is otherwise optional.
C. The problem is analyzed to determine the required qualitative change.
SIT’s solution stage comprises the aforementioned five idea-provoking operators. The five operators are divided into two groups of operators: those resulting in an extension of the functionality of the system (Unification, Multiplication), and those resulting in a restructuring of the system without adding new functionality (Division, Breaking Symmetry and Object Removal). We call each these two groups “solution strategies”. The first is called “Extension strategy”, and the second ”Restructuring strategy”. See Figure 4.1 for a flowchart of the SIT method.
Figure 4-1. A flow chart of the SIT Method(pp. 78-79)
4.1.1. The Preparation Stage
In the preparation stage, the problem solver is guided to collecting and organizing information about the given engineering system, its structure, its neighborhood and its associated undesired effects. The preparation stage consists of three consecutive steps described below (The first two steps are related to the given system itself, the third to the undesired effects associated with the system).
∙The problem solver determines the problem’s ‘world’ by forming a list of system and neighborhood objects.
∙The functional interrelations among system objects and their underlying technological concepts are determined through constructing a hierarchical Functional Model of the system. The functional model helps the problem solver recognize the system’s design rationale in general, and how each object in the system operates in particular. In the course of constructing a functional model, the problem solver may discover that one or more system objects are no longer necessary. Understanding the functioning of each object in the system is important in case a particular object is removed from the system and its role fulfilled by another. The functional structure is not a unique construct of the SIT method, and is used elsewhere as well (e.g. [Goel, 1997]).
∙Identifying as many Problem-Characteristics (see chapter 3 for definition) as possible. The problem-solving goal is determined as qualitatively changing at least one of them.
After carrying out these three steps, the problem solver is familiar with the design rationale of the system, with its neighborhood, and with various aspects of the relevant undesired effects. Both problem-solving goal and constraints are now determined in terms of the sufficient conditions. At this point the SIT method guides the problem solver to begin searching for solutions.
4.1.2. The Solution Stage
This section explains in detail SIT’s two solutions strategies: Extension and Restructuring and SIT’s Five Idea Provoking operators: Unification, Multiplication, Division, Breaking Symmetry, and Object Removal.
4.1.2.1. Solution Strategies
Any engineering system can be represented as a structure - a collection of interrelated physical objects and their relevant attributes, that support function - an operation that changes attributes of physical objects (See Goel, 1997). Consequently, when an engineering system suffers from undesired effects, the problem solver may begin the problem solving process by focusing on either the function or the structure of the system. Beginning the problem-solving process with a focus on functions, the problem solver thinks first of a possible new function that can reduce or eliminate any of the undesired effects (bring about a qualitative change in SIT terms). Since the Closed World condition confines the solution to only modifying (or eliminating) existing objects the new function must be later associated with an existing object (or at least an existing type of object). If the problem solver opts to begin the problem solving process in a focus on structure, he or she should first think of a modification of the existing structure (again, the CW condition confines them) and later verify whether that modification indeed results in a desired qualitative change. These two problem solving approaches, namely beginning with function and beginning with structure are called solution strategies in SIT. They are defined as follows:
∙1st. Extension strategy: first extend the functionality of the system by an addition of a new function (operation) that can bring about a desired qualitative change, then identify a closed-world object to carry out this operation
∙2nd. Restructuring strategy: first modify the system’s structure, than verify whether the modified system satisfies the qualitative change condition.
It is interesting to note that the restructuring strategy gives rise to a reversed thinking process, in which the structure of the solution is determined prior to understanding its meaning (or function). This reversed thinking process is defined by Finke [Finke, 1992] as a cognitive process of function follows form (FFF). As mentioned in Chapter 2, a series of experiments Finke conducted have shown that when people follow this process they produce more creative results. The second strategy then, forces the problem solver into an FFF thinking process. (pp. 79-82)
4.2.2. The SIT method Presented as An Interactive Computer Program
We now use the notation described below to present SIT in pseudo-code as an interactive computer program.
A. Preparation Stage
system objects list
[system object list, ,]
Neighborhood Objects List
[neighborhood object list, ,]
Functional Structure
[<system object list & neighborhood object list>@acceptors needs <system object list & neighborhood object list>@actors to directly perform on it a desired operation: [operation]@operations which is carried out according to the concept:
[concept]@concepts. This is {the primary | not the primary} function of the
^current object1^ in the system, ,]
Problem Characteristics
[undesired effects list, ,]
[Increasing the value of the attribute[Parameter]@causes increases the level of the undesired effect <undesired effects list>@effects, ,]
B. Solution stage
Strategy Selection
{→extension | →restructuring}
Extension
Conceptual Solution
[The relation <problem characteristics list>@prob_char will change from an increasing relation to {decreasing | unchanging}, if the following operation: [simple operation]@Simple Operation is performed, ,].
Extension technique selection
{→unification | →multiplication}
Restructuring technique selection
{→division | →breaking symmetry | →removing an object}
Extension Operators:
Unification
The object <system object list & neighborhood object list>@Selected Object will carry out the operation ^Simple Operation^. To do this, the object must be modified in the following way: [how the object will be modified].
Multiplication
New object(s) of the same type as <system object list & neighborhood object list>@Selected Object will be added to the system. The new object(s) will carry out the operation ^simple operation^. To do this, the new object(s) must be different from the original ^Selected Object^ in the following way: [In what way the new object(s) are different from the original one].
Restructuring Operators:
Division
The object <system object list & neighborhood object list> will be divided into {its basic parts | smaller elements of the same type | randomly}. A new degree of freedom will be achieved by locating each part in a {different place | different orientation | [other difference]}
Breaking Symmetry
Select an object <system object list & neighborhood object list>@Selected Object Form a list of important object parameters [list of parameters of that object]. The object ^Selected Object^ will be modified so that the object’s parameter <list of parameters> which is currently unrelated to the objects parameter <list of parameters> will be related to it in the following way: {increasing function | decreasing function | [other]}
Object Removal
The object <Object1s>@Removed Object will be removed from the system. Its operation ^the operation of the removed object^ targeted at the object ^ the target object of the removed object^ {→will be carried out by another closed world object |→will not be carried out any more, and to compensate, the system will be restructured}
Will be carried out by another closed world object
The closed world object that will carry out the removed object’s function will {continue | not continue}@Choice to apply the concept: ^concept of the removed object^
Idea provoking technique selection
{→unification (object removal) | →multiplication (object removal)}
Will not be carried out any more, and to compensate, the system will be restructured
{→division (object removal) | → breaking symmetry (object removal)}
Unification (Object Removal)
The operation, ^operation of the removed object^ targeted at the object ^target of the removed object^, that was carried out by the removed object, ^removed object^, will now be carried out by the object <system object list & neighborhood object list>@Selected Object. The new object will {continue | not continue} to operate according to the concept, ^the concept of the removed object^. To do this, the object ^ Selected Object^ must be modified in the following way: [how the object will be modified].
Multiplication (Object Removal)
The operation, ^operation of the removed object^ targeted at the object ^target of the removed object^, that was carried out by the removed object, ^removed object^, will now be carried out by new object(s) of the same type as <system object list & neighborhood object list>@Selected Object that will be added to the system. The new object will {continue | not continue} to operate according to the concept: ^the concept of the removed object^. To do this the new object(s) must be different from the original ^ Selected Object^ in the following way [In what way the new object(s) are different from the original one].
Division (Object Removal)
The object ^removed object^ will be removed from the system. Its operation ^operation of the removed object^ targeted at the object ^target of the removed object^ will not be carried out any more, and to compensate, the system will be restructured in the following way: the object <system object list & neighborhood object list> will be divided {into its basic parts | into smaller elements of the same type | randomly}. A new degree of freedom will be achieved by making each object different in {location | orientation | property [other]}
breaking symmetry (Object Removal)
Select an object < system object list & neighborhood object list>@Selected Object
Form a list important object parameters [list of parameters]Nparameters.
The object ^removed object^ will be removed from the system. Its operation ^operation of the removed object^ targeted at the object ^target of the removed object^ will not be carried out any more, and to compensate, the system will be restructured in the following way: The object ^Selected Object^ will be modified so that the object’s parameter <list of parameters> which is currently unrelated to the objects parameter <list of parameters> will be related to it in the following way:
{increasing function | decreasing function | [other]}
(pp. 85-89)
4.1.2.2. Idea Provoking Operators
Unification directs the problem solver to finding an existing system or neighborhood object to carry out the required operation. The object may be modified to adapt to the additional/new task, but must remain of the same type to satisfy the closed world condition.
Multiplication directs the problem solver to search for a new, slightly modified version, of an existing system or neighborhood object to carry out the required operation. (Note that adding new instances of existing types of objects is allowed under the closed world condition).
Division directs the problem solver to select one of the objects that belong to the problem’s world, break it down into its parts, and then reorganize the parts in space or in time.
Breaking Symmetry directs the problem solver to search for current symmetries (symmetries - in general, not limited to geometry or shape) and to try to recognize new states by breaking them. A symmetry is defined as a pair of unrelated variables (e.g. a circle is symmetrical in terms of angle – first attribute, and distance from its center to its circumference – second attribute). Breaking symmetry thus means connecting two hitherto unrelated attributes.
Object Removal directs the problem solver to remove an object from the system and then search for alternative (Closed World) objects to assume the function of the removed object (if necessary), or to restructure the system so that the operation, carried out by the removed object, will not be needed any more. The search for an alternative closed world object is performed with the direction of either Unification or Multiplication, whereas restructuring the system is conceived with the direction of Division or Breaking Symmetry (and even by Object Removal once again). (pp. 83-84)
4.6. Summary and conclusions
Although the SIT process is triggered by a description of an engineering system and its associated undesired effects, it can work even when no undesired effect is known, or when no system is given. If no undesired effect is known, it is possible to artificially generate one by hypothetically increasing the system’s performance until undesired effects begin to emerge. If no system is given and only an undesired effect is stated, the closed world condition would allow the creation of a system from objects that naturally reside where the undesired effect has emerged.
SIT’s output is a partial solution concept and not a full-fledged solution to a problem. The problem solver remains responsible for elaborating the information supplied by SIT into a detailed solution. SIT, like other known creativity enhancement methods, does not guarantee that all solutions that satisfy the conditions can be derived by the method. Solutions that do not incorporate any of the five idea provoking operators may be missed. Three sets of creative solutions can be defined: solutions derivable by the method ⊆ solutions that satisfy the conditions ⊆ all creative solutions.
Like any other problem-solving aid SIT is not free from limitations and this should be stated explicitly. In order to compile the necessary information from the problem statement into the language of the conditions, the mechanism of the problem itself must be relatively well understood, and the problem must be reasonably well delineated in space and time. When the situation is highly complex, ambiguous, and not confined in space and time SIT looses much of its effectiveness. In such cases it is hard to identify both the objects that comprise the relevant closed world and the required qualitative changes. Another limitation is related to the fact that SIT is geared to support the generation of a specific type of solutions, those satisfying the conditions. Obviously, there may be other solutions. Problem solvers must be aware of this fact, especially if SIT fails to produce satisfactory results. (pp. 113-114)
6.1. Product, process, and person
Chapter 3 studied the unique characteristics of creative engineering products, as opposed to conventional or routine ones. The results took the form of two jointly sufficient conditions that characterize a large set of creative engineering solutions that fall into the category of ‘inventions within a paradigm’. The first condition - the Closed World condition - constrains the solution to not incorporating new types of object that were absent from the initial engineering system and its neighborhood. The second condition - Qualitative Change - requires that the solution incorporate a qualitative change in at least a single relation (between an undesired effect, and an attribute that currently increases the severity of the undesired effect) that characterizes the problem. “Qualitative” means here that a relation changes its trend from a direct relation to either no-relation or inverse-relation. Setting problem-solving goal in terms of relations rather than absolute values requires major changes in the behavior of the system. This, in conjunction with the constraints posed by the Closed World condition, creates a problem-solving task solvable only by creative operators.
The five idea-provoking operators are divided into two main strategies that reflect two different cognitive approaches to problem-solving. In the first, the extension strategy, a top-down process is applied in which the problem solver first determines what to do and then how to do it. With the second approach, the restructuring strategy, a bottom-up, function-follows-form process is applied in which the system is first restructured (by modifying the interrelations among its parts or attributes) followed by an attempt to assign meaning to the new structure. In the restructuring strategy, the how precedes the what.
In its initial stages SIT guides the problem solver in analyzing the problem’s world: to prepare a list of system and neighborhood objects; to determine the functional relations among system objects; and to identify the problem characteristic relations. Empirical pre-training and post-training studies demonstrated that, indeed, the rate of conditions-satisfying solutions increased drastically through SIT training and application. (pp. 142-143)
6.2. Theories of Creativity Revisited
Table 6-1. Summary of the relations between other theories’ concepts and the concepts of this work
(pp. 144-147)
Chapter 2 ended with the following paragraph that suggested the possibility of selecting and integrating different theories of creativity in order to arrive at a concrete set of instructions for directing problem solving towards creative solutions.
Set the problem solving goal as overcoming a contradiction or second order change, but confine your search-space to variations on a current theme (rather than abandoning current theme and moving to another one). This may lead you to a temporary state of failure which is a good starting point for the creative process. Draw on your divergent thinking ability and skill, mainly fluency and flexibility, to expand the search-space. Be aware of the possibility that the solution lies in an isolated area of the search space due to functional and structural fixedness. Use heuristics such as reversal, dropping constraints, function-sharing, and changing the value of a numeral.
Using this work’s terms and concepts, this paragraph can be translated into the following set of instructions showing the close correspondence between our theory and (an integration of many) previous ones:
Set the problem solving goal as qualitatively changing at least one problem characteristic unit, but confine your search-space to the current Closed World. This may lead you to a temporary state of failure which is a good starting point for the creative process. Explore SIT’s strategies and techniques: Unification, Multiplication, Division, Breaking Symmetry, and Objet Removal, focus on different objects, and try to view each from the viewpoint of different meaning
dimensions. (pp. 147-148)
6.3.1. Non-Closed-World Creative Engineering Design
The framework of the sufficient conditions characterizes a large set of creative engineering solutions but not all. Creative engineering solutions exist in which new types of objects are added to the system. More research is needed for characterizing such solutions. One hypothesis may be that in creative solutions, when incorporating non-closed world objects, the new object forms a special type of interaction - yet to be discovered - with the given system.
Another type of engineering problem for which the framework of the sufficient conditions is currently not inapplicable is design-from-scratch problems in which the problem’s world is simply not yet defined. Great inventions such as the telephone, the laser, the light bulb fall into this category. Although we believe such inventions stem from different creative mechanisms − dependent on the development of basic technologies, chance, spirit, and large-scale social trends − it would be interesting to look for common patterns for these types of engineering problems as well. (pp. 148-149)
6.3.5. New Domains
Most of the presented examples dealt with problems of a physical nature. A future research topic may be the extension of the theory to other fields, such as management, business strategic planning, marketing, advertising, and new product development. The main difficulty in applying the framework of the sufficient conditions to these new domains will be the formulation of the Closed World condition. The challenge will be to replace the notion of physical objects, which currently constitute the core of the closed world, to other, more abstract constructs. (p. 150)
감상:
너무도 아름답고 훌륭한 논문임...... 발표된 후 22년이 지난 지금 보게 된 것 또한 시의적절하다. 8월 14일부터 읽기 시작해서 내용을 음미하는데 오늘까지 11일이 걸렸다. 그간의 시간은 즐거움의 향연이었다.
(2021. 8. 24)
