INTERNATIONAL CONFERENCE ON ENGINEERING DESIGN
ICED 05 MELBOURNE, AUGUST 15-18, 2005
Morphological Analysis (MA) leading to Innovative Mechanical Design
John Dartnall* email: firstname.lastname@example.org and
Stephen Johnston email: email@example.com
Keywords: Morphological Analysis, design space, innovation, creative design
In this paper we apply Morphological Analysis (MA) to a conceptual design where the
designer is attempting to exhaustively search a manageably defined design space in order to
select the most appropriate (innovative) design solution.
The morphological approach has not yet been fully developed or extensively applied. A
common view is that MA is simply the “morphological box” as it is sometimes called. In our
view morphological analysis is a methodology to be applied creatively, adapting and
modifying it to meet the requirements of the problem on hand. The essence of the
methodology is that the designer should be looking for new opportunities and for
contradictions, and be prepared to modify a previously defined exhaustive morphological
procedure as new information surfaces.
In our case study the initial search was to find all the ways to assemble the basic elements of
down-hole water lifting piston pump. As we adaptively applied MA, we recognised that there
were more sophisticated assemblies of the basic elements and we generated the down-hole
pump patents of over 100 years as well as some configurations that were not found in the
2 Background and Literature about Morphological Analysis
Many of the textbooks on engineering design process give a basic introduction to what earlier
advocates of the morphological method proposed (Ullman, 2003; Otto and Wood, 2001;
Cross, 2000; Dym and Little, 2000; Wright, 1998). These texts present it as a creative method
for generating alternative solutions to a design problem. They illustrate the method by
describing an example or two where a list of functional requirements (FRs), sub-problems or
concept sub-modules are listed on one axis (say the vertical axis) of a matrix whilst variants
satisfying each of these FRs, sub-problems or sub-modules are entered along the other axis.
The designer then trials design solutions comprising various combinations, judiciously chosen,
constructing them by taking one element from each row.
Some authors (Pahl and Bietz, 1996; Hubka and Eder, 2000; Dieter, 2000; Roozenburg an
Eekels, 995) offer a broader view. French, (1999) has an entire chapter, Combinative Ideas,
on this area. French gives a range of interesting examples demonstrating some broader
applications of morphological method, unfortunately without mentioning the word
The word morphology derives from the Greek word morphé meaning “form”. Plato and
Aristotle effectively employed morphological principles in their descriptions of the animal
kingdom. Goethe, in the late eighteenth century, described the structural interrelationships
between things as morphology. According to the Darwinian biological evolution theory,
continued improvement of species results from advantageous genetic mutations. This theory
suggests a model for the processes of technological design, where morphological examination
of designs can identify superior mutations.
“Morphological analysis performs a systematic exhaustive categorization and evaluation of
the possible alternative combinations of sub-capabilities which may be integrated to provide a
given functional capability”, (Martin, 1994).
One of the best known examples of a morphological box is the Chemical (Mendaleev)
Periodic Table. The orderliness of this table (increasing atomic number and shell completion)
enabled the discovery of “missing elements” and facilitates the learning of the properties of
the families of elements.
Zwicky (1969) is widely cited by design process authors in connection with morphological
method. His goals were to expand the search space and to safeguard against overlooking good
novel solutions to a design problem. Zwicky had a deeper view underlying his proposal of
morphological analysis as a basis for discovery, invention and research. He proposed that all
facts should be thoroughly investigated and properly appraised for the purpose of selecting
the things among them that best satisfy our requirements. Zwicky states:
“the morphologist must never lose sight of the continuity of all things, all phenomena and all
concepts and all mental outlooks… nothing should be discarded as unimportant… the
morphologist will persevere where others have long since given up the effort”.
Zwicky also gives examples of different classes of morphological method. His first category
is described as “Systematic Field Coverage” by which, as an example of the method, he
generates all five possible regular polyhedra: tetrahedron, octahedron, icosahedron, cube and
dodecahedron, finally demonstrating the impossibility of constructing regular polyhedra from
regular polygons having more sides than the pentagon by demonstrating that the hexagon tiles
a flat plane.
In Zwicky’s second category he introduces the morphological box (matrix) and in
exemplifying this category gives a 10 X 10 matrix of energy transformations where a
represent energy transformations from one form to another (eg. elastic energy into kinetic
energy). Under each energy conversion heading he then outlines historical information and
examples of that particular type of energy conversion.
Zwicky also illustrated third, and final category, is entitled “The Method of Negation and
Construction” which he employed in modifying Newton’s Universal Law of Gravitation
where distances between galaxies are greater than about 100 million light years. At these
distances, Zwicky’s investigations indicate that the attraction between mutual bodies is much
smaller that the classical inverse square law predicts.
Zwicky gave a number of examples from his own morphological work on rocket propulsion
systems for which he was awarded numerous patents.
3 Morphological Analysis and Combinatorial Explosion
It is well documented that morphological methods may lead to the problem of combinatorial
explosion. Various techniques have been proposed to deal with this problem. In discussing
this problem as it relates to the common method where the morphological matrix has FRs in
the rows and solutions in the columns, some authors suggest evaluating the individual
solutions, row by row, and then combining the optimum row ones to obtain the overall
solution (Pahl and Beitz, 1996; Al-Salka and Cartmell, 1998). This will reduce the number of
evaluations from a power relationship (m
) to a multiplicative relationship (m*n) in a
situation where the number of FRs is n and the number of solutions to each FR is m.
Mathematicians, in solving problems involving large matrices (before the advent of the
computer) developed methods of partitioning the matrices, so that they could work on the
individual partitions in a more manageable way. When setting up matrices to model
engineering situations, the viability and value of partitioning is often readily apparent from
the physical situation, as for example when applying finite element analysis to a stress
In designing/inventing an engineering system, the problem will most likely suggest some
manner of breakdown into subsystems/sub-problems. Morphological method may then be
applied to each of these. One must bear in mind that there are likely to be interactions
between the sub-systems and that both the overall problem and the sub-problems are dynamic
throughout the design process.
The morphological approach outlined by Zwicky has not yet been developed and extensively
applied at the rigorous level that he believed in. His idea was not simply the “morphological
box” as it is sometimes called. He saw it as a methodology, not merely a method. He intended
it to be applied creatively and modified to suit the problem on hand. To sum up this point, the
designer should be prepared to look for new opportunities and for contradictions, while
pursuing a previously set exhaustive procedure.
4 The Need for Rapid Sketching and Recording Ideas during the
Several researchers such as Ullman have carried out protocol studies of engineering designers
at work, doing design, in order to produce cognitive models of the design process. (Ullman et.
al, 1988, 1995 and 1997).
Ullman and his co-workers were concerned that computer based systems should be developed
to assist designers and design teams in their work. They observed that, as computers increased
in capacity and speed, more powerful design software such as FEA packages and solid
modelling packages were developed. However these packages, particularly the earlier ones,
were time consuming and absorbed a great deal of the users’ energy in mastering their use. At
the same time, they were rather specialised into particular areas such as detail drafting and
various particular design analyses. Many of them did not easily allow the designer to work
seamlessly between the processes involved. They also lacked facilities for recording the
history of design process, the decisions made during the design process, and the relevant
criteria. We will now discuss some of the goals of the 1995 paper by Ullman, which was
directed toward the idea of an “Ideal Mechanical Engineering Design Support System” These
goals are best related to Figure 1.
Ullman defines the term "architecture" as “the stick figure that can be easily manipulated and
changed before the shape is refined. Shape implies the geometry that adds body and detail to
the architecture. Often designers first develop the general architecture of the object being
designed and then they add details about shape and fit.” Ullman’s diagram shows these
entities in a central position. Form refers to both architecture and shape.
The short term memories (STM) of the designers act quickly as the architecture comes to
mind whilst they attempt to meet functional requirements and constraints related to parts and
sub-assemblies. The computer software should support the graphical documentation of this
architecture in a speedy and non-burdensome way and allow recording of reasons behind the
ideas and decisions. Function happens primarily at interfaces between components making up
Figure 1: Ullman’s diagram related to design decisions
As the design progresses (moving outwards in Figure 1), information related to manufacturing,
assembly, materials and cost of the current concepts need to be to be readily accessed. The
ideal computer software should support this access so that rapid evaluation along with
documentation can take place.
Next, Ullman points out that as the design moves from conceptual to layout to detail,
constraints and limitations come to light in addition to the initial functional requirements
(FRs) that may have resulted from a QFD type process. Ullman refers to both initial FRs and
evolving constraints as requirements and he recommends that the software integrate the
management of these requirements into the development of parts and assemblies.
The term “issues and plans” includes ideas such as Integrated Product and Process Planning
(IPPD), the successor to concurrent engineering. Again, the computer software should support
issues and plans.
What does Ullman mean by Design Intent? He explains that the ideal engineering design
support system should manage all the items inside Figure 1 in a database. In addition it should
support detailed information about arguments for and against decisions made based on
requirements as well as recording the decisions themselves. He points out that other authors
have used terms like design history, rationale and corporate memory in a similar vein to his
expression of "design intent".
Studies of human information processing have shown that over two thirds of the strategies
used by design engineers whilst developing new products were searches through the design
space. The artificial intelligence community has put much effort into the area of search
strategies. Efficiency in capturing, archiving and querying the full range of design information
is clearly an important goal for a design support system.
When the first author started to generate the morphology of down-hole piston pumps, he
initially employed two techniques (Dartnall, 2003):
• The use of an Excel spread sheet and careful symbolic labelling of all combinations
and permutations to represent the different pumps.
• The construction of solid models of each pump.
Both techniques were tedious. Visualisation from the first technique was so difficult that
eventually images would have to be constructed in order to present the results for the benefit
for communication to others. Construction of the solid models was time-consuming – taking
times ranging from about 1 hour to one day for each model. He realised early that there were
in excess of 60 models involving a large amount of tedious work. Clearly, some form of stick
diagram is desirable, as indicated by Ullman.
However the designer needs to do a lot more than the tedious construction of solid models or
even the more expedient construction of stick diagrams. The designer needs to be evaluating
each idea, possibly each step of the construction to see how it affects various performance
requirements, cost, design for manufacture (DFM), design for maintenance, design for
assembly (DFA), efficiency, life cycle, reliability, how it supports the total concept. Manual
sketching, in the past has allowed designers, as individuals or in groups, to discuss these
issues over quick sketches.
5 Morphology of Down-hole Piston Pumps – Case Study
We now show how, an initial morphological analysis was attempted by constructing a matrix
(spread sheet of Table 1) having the basic piston pump elements in columns and the various
conceptual pumps, systematically constructed in the rows. It was soon recognised that the
pumps generated could be categorised into categories such as Disc-piston pump and Tube-
piston pump. Many such categories were anticipated and these were added to the summarising
Table 2 as they were recognised. Some of these were recognised later in the process.
Some of the solid models of the early conceptual pumps are shown in Figure 2. The
constuction of these was time-consuming and led to the idea of rapid graphical generation of
conceptual pumps from elements of the code of Figure 3.
The rapidly constructed computer sketches were constructed in Microsoft Word, after
defining the graphical code of the basic elements such as valves, valve elements such as discs
and cages, tubes, seals etc., that are common to all conceptual piston pumps. The various
down-hole pump configurations were then built from these by copy and paste operations. In
this way, the rapid computer sketching of conceptual pumps was not burdensome.
Table 1: Combinatorial generation of some early symbolic models of down-hole pump configurations
(morphological analysis of the first category, 10.00).
Figure 2: Some early solid models constructed for generation of down-hole pump configurations
By colour coding and employing conventions such as drawing only half of any symmetrical
(completely round) pump, the process was visually efficient. The entire pump was drawn if it
contained any non-symmetrical feature. Symmetry is both a DFM/DFA issue and a functional
issue in the down-hole pump design. From the DFM/DFA point of view, symmetry implies
faster and less expensive machining operations such as turning and drawing as well as rapid,
screwed assembly. From a functional view, round, symmetrical elements, such as tubes and
rods can be concentrically assembled and are helpful in minimising the diameter. A minimum
diameter design to fit inside a minimum diameter well is highly desirable. Screwed assembly
also has the advantage of being readily disassembled for maintenance.
Another evaluating criterion of the designs is the problem of abrasive particles being captured
in the seals and wearing the mating walls. This problem needed to be thought through with
every pump configuration. No “dirt pockets” are permitted, especially above seals, as the dirt
is sure to cause serious wear.
A further desirable feature of a pump design is to minimise leakage from the seals and valves,
not only during operation but also during periodic non-pumping periods such as when there is
a lull in the wind to a wind-pump. This often leads to the leak-back of the entire delivery
column during a lull. In order to achieve minimal long term leak-back, valve and seal design
and positioning are very important. The designer will have many of these sorts of problems to
check out as he/she works through the different configurations of a design.
The first five of the pumps generated from the graphical code are given in Figure 4. In the
first pump, water is delivered via an eccentric pipe containing the delivery valve. This pump
is drawn in full, whereas the other four pumps are comprised of concentric circular elements
and therefore only their left half is drawn. The first pump, although once widely used in the
Watt era, is not suitable for modern small diameter tube wells.
Some comments on the remaining pumps of Figure 4 are now given. They demonstrate how
evaluation of each concept may take place.
Pumps no 2 and 3 of Figure 4 are widely used in conventional wind-pumps. These devices
have had a long history of premature wear when used to pump water containing abrasive
sediments as these sediments tend to become captured between the lip seal and the cylinder –
i.e. a "dirt pocket" effect. In addition, friction tends to axially compress the lip of their seal
especially during high pressure surges that occur due to water column inertia at the beginning
of a pumping stroke. It is common knowledge that excessive wear is found at the bottom of
wind-pump cylinders where inertial pressure pulses have their effect.
Pumps no 4 and 5 of Figure 4 have inverted seals, external to the water flow in order to avoid
both the dirt pocket effects and the axial lip compression effects present in the above two
pumps. However they are single volume even though pump 5 is double acting. In addition,
because a tube containing the delivered water is reciprocated rather that a rod they have
further complications at their interface with other sub-modules (such as the delivery pipe) of
the complete pumping system.
As generation of further conceptual pumps took place, the thought occurred to the author to
investigate the idea of having two moving valves, rather that one fixed and one moving. This
resulted in the configuration 12 of Figure 5. It had certain advantages such as smooth flow, as
outlined in the diagram text. However the advantages were obtained at the expense of the
extra complexity of having to drive each piston valve separately.
Figure 3: Code for rapid graphical generation of conceptual down-hole pumps
Figure 4: Some of the early conceptual pumps generated by the rapid graphical technique. Note the numbers do
not match the categories of Table 1. They are respectively 10.01, 10.03S, 10.03R, 10.08S and 10.08R.
Figure 5: Two innovative concepts, both involving two moving piston valves, one having independent piston
valve movement and one having the piston valves ganged together.
The next thought was to see if the above situation could be simplified by driving two piston
valves with one reciprocating rod. Figure 14 illustrates that this could be done by forming the
cylinder into a "U" tube. This configuration, of course is, is not suitable for fitting down tube
wells. Can this conceptual design be re-configured so that it is essentially constructed from
concentric circular elements?
Pump 14 can be rearranged by placing one leg of the U tube above the other, as shown in
pump 16 of Figure 6. Unfortunately, we still have a space wasting eccentricity, so we
transform the rod joining the two piston valves into a tube. Now we arrive at pump 17 in
which we have the flow path of pump 16 in a pump made up from essentially circular
This is one of the author's configurations. It
is a double acting (say, 1.75 acting) rod
driven pump. A potential advantage with
respect to maintenance is that both valves
and the seal are on the rod and could be
extracted from the well with the rod without
removing the cylinder and its casing. Slim,
This double acting tube well pump was
"invented" by the author. It is a conceptual
model Its layout is similar to the Deming
pump of the early 1900’s, one difference
being that a piston valve replaces the lower
piston of the Deming pump. Another is
that the intake is through the open top of
the bottom cylinder. The bottom intake
valve of the Deming pump is deleted. The
author's next pump is a nested
transformation, which follows from the
circuit of this pump
Figure 6: Pump 16 is an intermediate transformation from the conceptual pump 14 of figure 5, leading the
practical double acting pump 17
Many configurations were generated and it was important to evaluate the configurations in
conjunction with other modules of the pumping system.
As superior deigns emerged it was useful to construct solid models of these in order to
evaluate them more thoroughly. An example is shown in Figure 7.
Transfer passage from
Fluid delivery into outer casing
Figure 7: The conceptual solid model of a pump that is essentially an inversion of pump 17. The flow
between the bottom and the top stages is via an external annular passage. Another difference is that water
enters through a bottom valve rather that a side port.
6 What are the benefits of the kind of Morphological Analysis of
• The procedure demonstrated in our case study has revealed the following:
• In order search for improvements in a device such as the down-hole wind pump it is
highly desirable to be able to rapidly construct the many conceptual designs that are
• The potential techniques to aid the search include matrix based analytical techniques
and various graphical techniques. Rapid graphical techniques greatly reduce the time
wasted on meaningless and non-functional combinations that will nearly always be
present when a complete combinatorial analysis is carried out.
• Rapid graphical techniques enable immediate evaluation of the generated conceptual
designs to be made.
• Rapid graphical techniques along side the appropriate analytical matrix enable the
researcher to see where new opportunities are.
• This case study supports Zwicky’s claim about the power of morphological method in
that it rapidly generated numerous conceptual designs. An indication of this power is
that the conceptual designs that it generated were the patents of the past hundred or so
years in this area.
• Software designed and constructed on the basis of the techniques outlined in this case
study would be useful.
By way of a case study, this paper has demonstrated the use of morphological analysis in
innovative mechanical design. It has demonstrated that there is more to morphological
analysis than the commonly described “morphological box”. Designers need rapid sketching
facilities to assist them as they progress from one idea to another, making evaluations at each
stage. These sketches need to reveal sufficient information to assist the designers in their
assessment and decision making processes without being burdensome. The process needs to
facilitate the type of thinking that will enable the designer to see new combinations of
elements and new categories of the morphological analysis. The case study demonstrated
some foundations (efficient combination of tabular and graphical procedures) from which
useful software could be produced to assist the designer in more effectively applying
[1.] Al-Salka, M. A., Cartmell, M. P. and Hardy, S. J., 1998; "A Framework for a
Generalized Computer-based Support Environment for Conceptual Engineering
Design", Carfax Publishing Ltd.
[2.] Cross, N., 2001; "Engineering Design Methods: Strategies for Product Design (3nd
ed)", New York: John Wiley & Sons.
[3.] Dartnall, W. J., 2003; "Innovative Mechanical Design with a Case Study of Pumping
Systems for Low Yield Tube Wells", ME (thesis), University of Technology Sydney.
[4.] Dartnall, W. J., 2004 and Johnston S. J; "Trend-Morph-Pds, A Methodology For
Innovative (Mechanical) Engineering Design", 7th Biennial ASME Conference on
Engineering Systems Design and Analysis
[5.] Dieter, G. E., 2000; "Engineering Design, 3rd Ed", McGraw Hill, N.Y.
[6.] Dym, C. L. and Little, P. 2000; "Engineering Design: a Project-Based Introduction",
John Wiley and Sons Inc; New York.
[7.] French, M., 1998; "Conceptual Design for Engineers (3rd Ed)", London; New York:
[8.] Hubka, V. and Eder, E. R., 1996; "Design Science: Introduction to the Needs, Scope
and Organization of Engineering Design Knowledge", London: Springer-Verlag.
[9.] Martin, M. J. C., 1994; "Managing Innovation and Entrepreneurship in Technology
Based Firms", John Wiley and Sons Inc; New York.
[10.] Otto, K. N. and Wood, K. L., 2001; "Product Design, techniques in Reverse
Engineering and New Product Development", Prentice Hall, New Jersey.
[11.] Pahl, G. and Beitz, W., 1996; "Engineering Design: A Systematic Approach (revised
2nd ed)", London: Springer-Verlag.
[12.] Roozenburg, N. F. M. and Eekels, J., 1995; "Product Design, Fundamentals and
Methods", Chichester: Wiley.
[13.] Ullman, D. G., Dietterich, T. G. and Stauffer, L. A., 1988; "A Model of the
Mechanical Design Process Based on Empirical Data", AI EDAM Journal vol. 2, no.
[14.] Ullman, D. G., 1995; "A Taxonomy for Classifying Engineering Decision Problems
and Support Systems", Artificial Intelligence for Engineering Design, Analysis and
Manufacturing, no. 9, pp. 427-438.
[15.] Ullman, D. G., et al., 1997; "What to do Next: Using the Problem Status to Determine
the Course of Action", Research in Engineering Design, 9, pp 214-227.
[16.] Ullman, D. G., 2003; "The Mechanical Design Process, 3rd ed.", McGraw Hill, New
[17.] Wright, I. C., 1998; "Design Methods in Engineering and Product Design", McGraw
[18.] Zwicky, F., 1969; "Discovery, Invention, Research through the Morphological
Approach", Macmillan, Canada.