Documente Academic
Documente Profesional
Documente Cultură
C
OMPANIES THAT DESIGN COMPLEX, highly engineered prod-
ucts all have their horror stories. Ford and Bridgestone
Firestone lost billions of dollars after their failure to co-
ordinate the vehicle design of the Ford Explorer with the
design of its tires. Similarly, Airbus’s development of the A380
“superjumbo” suffered major delays and cost overruns because
of late emerging incompatibilities in the design of the electri-
cal harnesses of various sections of the plane’s fuselage. These
mistakes probably contributed to the loss of Airbus’s CEO and to
important changes in the management of the A380 program.
What’s striking about these stories and many others like them
Jude Buffum
developing the products’ components management tool: the design structure gine was divided into eight subsystems,
had communicated more effectively. Of matrix (DSM). Our application helps each of which was further decomposed
course with complex development proj- managers identify where failures in into five to ten components, for a total
ects, you can never be certain that you planned communications could occur of 54 engine components. Typical for
have planned for every contingency. as well as recognize when project teams such projects, the development organi-
However, our experience shows that engage in unplanned technical commu- zation was correspondingly structured
in the design phase of such projects, nications. We also analyze communica- into 54 cross-functional design teams
many companies would benefit from tions between project teams that take responsible for each component, plus
focusing sharply on the critical points place both within and outside the for- six integration teams responsible for
of contact among their various compo- mal project structure. We conclude by managing engine-level requirements
nent development teams to ensure that discussing how managers should han- in areas such as fuel efficiency. These
everyone knows when and with whom dle communications problems that are teams had to interact a lot: There were
they should be sharing information. revealed in the process. While we do not several hundred interfaces among the
This article helps managers mitigate pretend to offer a definitive solution engine components, many of which
this design communication problem. to the design communication problem, would have experienced significant
Drawing on detailed research into how we do believe that managers who use problems without proper communica-
Pratt & Whitney handled the develop- our tools over succeeding generations tion among the relevant design teams.
ment of its highest-thrust commercial of products will improve the quality of To help manage the communications
jet engine – the PW4098, which pow- their development processes. aspect of such projects, we propose the
ers the Boeing 777 – we present a new following approach: (a) identify unat-
application of an established project Catching Missed Interfaces tended interfaces, areas where commu-
Before They Occur nication should be occurring but is not,
Article at a Glance The first thing a project team does when and (b) look for unidentified interfaces,
faced with a complex development areas where communication is occur-
Companies that design complex prod- challenge is break the project down ring but has not been planned. The re-
ucts all have their horror stories. Yet, into manageable pieces that are then sult of implementing this approach is
they can all avoid mistakes by ensuring assigned to dedicated subteams. In the what we call an alignment matrix, which
that the different teams responsible for
context of developing a product like a reveals mismatches between the com-
developing the components of the prod-
jet engine, this results in a large number munications and exchanges that are
ucts communicate more effectively.
of specialized cross-functional teams, supposed to occur and those that actu-
A new application of an established each working on a different component ally do. It also demonstrates, therefore,
project management tool, the design or subsystem of the engine. Of course, how well the project has been planned
structure matrix, can help a company these teams cannot work in isolation; and executed.
identify where failures in planned in addition to designing their assigned To see how the approach works,
communications could occur as well as components, they must also integrate let’s suppose that we plan to develop
recognize when project teams engage their designs with those of the other a product with four components, each
in useful technical communications that
components to ensure that the entire of which requires its own specialized
were not planned.
product or system functions as a whole. design team. (This approach may be
If a company finds that a lot of planned It is critical, therefore, in planning a used when the organizational structure
communication is not taking place, then complex product development process maps directly to the product architec-
it should revisit its product develop- that the project managers specify just ture – that is, component X is designed
ment organization. Even projects that which resources and information differ- by team X – which is the case in most
are completely organized around a ent teams will need from each other at of the complex development projects
product’s architecture are typically vul- particular stages of the project. in the automobile and aerospace firms
nerable to communication breakdowns. In the Pratt & Whitney jet engine we have studied. For cases in which
development project we studied, the en- the organizational and product struc-
A company can ensure that critical com-
munication occurs by tasking special
teams (or the teams involved) with Manuel E. Sosa is an assistant professor of technology and operations management at
making sure that the right people talk to Insead in Fontainebleau, France. Steven D. Eppinger is a professor of management science
each other. It’s also important to ensure and engineering systems and deputy dean at MIT’s Sloan School of Management in Cam-
that the teams are working with the bridge, Massachusetts. Craig M. Rowles was a Module Center manager at Pratt & Whitney,
optimum communication tools. based in East Hartford, Connecticut; he is now CEO of Emegear, a medical device company
in Carpinteria, California.
components
D • A B C D sponding team interaction
A • Unidentified interface: team
B • interaction that takes place
Team Interaction Matrix or is expected to take place
Providing teams C • without a corresponding design
interface identified by system
Design A B C D D •
architects
Receiving teams
Teams’ A •
Input Lack of interdependence:
determine technical B • components that do not share
interactions teams an interface or involve design
C •
have had, are having, team interactions
or expect to have D •
Team C requests
information from Not applicable
team D •
tures do not map directly, refer to M.E. transfer forces, material, energy, or in- people working on each component
Sosa’s “Aligning Process, Product, and formation to other components to en- design team expect to take place with
Organizational Architectures in Soft- able them to work properly? Answers people from the other teams. Specifi-
ware Development” in the Proceedings to such questions are used to identify cally, we ask members of each team
of the 14th International Conference in the interfaces among all the compo- whether they anticipate the need for
Engineering Design, Paris, August 2007.) nents of the product. technical information or resources
Our analysis involves the following Armed with this data, the project from other teams. In surveying them,
three steps: managers can present the responses we need to make sure they are all fa-
1. Interview the system architects. on a four-by-four design interface ma- miliar with the function and specifica-
We begin by requiring the senior en- trix (a type of DSM used to map the tions of the component or components
gineers responsible for the product’s network of component interfaces), they are developing. (We do not share
overall function and layout – the sys- such as that illustrated in the exhibit with them the matrix produced in the
tem architects in engineering lingo – to “Assessing the Fit Between Design and first step, as this would bias the teams’
identify the technical design interfaces Communication.” responses.) Using these survey data, we
among the four components. Do com- 2. Survey the component design can represent the technical interaction
ponents need to be spatially connected teams. In the second step, we identify patterns of the teams in another four-
with each other? Do some components the technical communications that the by-four matrix (corresponding to the
four design teams) that we call a team state where they actually received the (vibration or heat), structural forces, or
interaction matrix, also shown in the ex- information and resources they needed. signals used by the control system of
hibit “Assessing the Fit Between Design Overlaying the new team interaction the engine. The design interface matrix
and Communication.” matrix on the original design inter- drawn from this information is shown
3. Combine the results. In the third face matrix would reveal whether the in the exhibit “Creating an Alignment
and final step, we overlay the two ma- mismatches uncovered at the start of Matrix for Pratt & Whitney.” We then
trices to obtain the alignment matrix the project had persisted and whether asked at least two members of each
(again, see the exhibit “Assessing the other mismatches had appeared. A of the 54 teams involved in the proj-
Fit Between Design and Communica- postmortem of this kind yields valuable ect how often they received technical
tion). This matrix shows the matches insights into future product develop- information from other teams during
and mismatches between the product’s ment projects, especially for companies the detailed design phase of the proj-
ect and how critical they perceived this
information to be. The results of this
survey documented 423 interactions
among component teams, which ap-
pear on the team interaction matrix
shown in the exhibit. We finally com-
puted an alignment matrix by merging
the two first matrices.
Readers armed with a magnifying
glass would count 220 unattended de-
sign interfaces that were not matched
by team interactions and 74 uniden-
tified interfaces in which teams ex-
changed technical information even
though there was no identified design
interface between the components. Al-
though it would be naive to expect a
perfect alignment of design interfaces
and team interactions – and, in this
case, many of the 220 unattended in-
terfaces were unproblematic or not
critical – misalignment on this scale
indicates that Pratt & Whitney was sub-
architecture as conceived by the system that expect to develop similar products ject to considerable risks involving cost
architects and the expectations of the in the future or further generations of overruns or other problems with this
teams involved in the product’s devel- the same product. project.
opment. More important, it highlights We conducted one such analysis of
the mismatches – when design inter- Pratt & Whitney’s development of the Why Mismatches Occur
faces have been identified but team PW4098, the engine that, at the time, Mismatches do not occur randomly
interactions are not taking place (un- set new standards in the aviation indus- in a product or organization. Rather,
attended interfaces) and when team try for development speed and cost. (It they are the result of product design
interactions take place without a corre- was also the first commercial jet engine and organizational factors. Planned
sponding design interface being identi- ever certified for 180-minute extended- key communication points may remain
fied by system architects (unidentified range twin operations from its first day problematic for several reasons, in-
interfaces). As we shall see, sometimes of service.) We began by interviewing cluding the presence of organizational
these unidentified interfaces turn out the engine’s architects, who identified boundaries (cross-boundary interfaces
to be critical. 569 interfaces among the 54 main com- are more likely to be missed than in-
At the end of a project, managers ponents of the engine. Many of these terfaces with a team belonging to the
can update the alignment matrix by interfaces were critical and complex be- same group), the lack of interface criti-
redrawing the team interaction matrix cause they not only involved physically cality (complex and critical interfaces
based on more recent surveys in which adjacent components but also the trans- receive more attention than noncritical
component team members are asked to fer of material (air, fuel, or oil), energy ones), the use of indirect communica-
2
3
4
8
6
6
4
4
2
1
2
0
0
0
0
4
0
0
4
0
4
4
0
4
0
6
0
4
6
0
0
0
6
6
0
0
4
4
0
0
0
0
0
0
8
8
0
0
6
6
5
5
2
2
5
5
2
2
2
2
0
0
0
0
0
0
569 design interfaces among
engine components
0 6 3 1 6 6 0 0 0 0 0 0 0 0 0 0 0 0 0
2 5 2 0 0 4 4 0 0 6 0 0 0 0 0 0 0 0 0 0 0
4 5 0 0 0 0 0 0 6 0 4 6 0 0 0 0 0 0 0 0
0 0 0 0 0 8 8 0 0 3 0 4 0 0 0 0 0 0 0 0
0
0
6
4
4
0
6
0
0
0
0
3
4
0
0
5
0
0
7
0
4
0
2
7
0
0
3
3
6
0
7 0
4
5
0
0
3
3
0
3
6
3
6
6
0
0
6
2
2
2
4
0
0
2
2
2
2
0
4
0
0
0
3
2
2
0
4
0
0
0
0
0
0
0
1
0
0
0
0
0
1
0
0
1
0
0
0
0
0
1
0
an actual team interaction
0 6 0 3 0 0 0 0 0 0 4 0 0 2 0 2 2 4 0 0 4 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 6 0 0 0 0 2 4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 2 0 0 6 0 0 0 0 0 2 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 6 0 0 0 0 2 2 0 0 0 0 0 3 2 0 4 0 0 0 0 1 0 1 1 1 1 0
0 6 0 2 0 0 0 0 0 2 2 2 2 2 2 2 0 4 0 2 4 2 0 0 2 0 0 0 0 0 0 0 1 1 1 1 1 0 0
3
0
3
0
3
6
3
3
0 0
0
0
0
0
2
0
0
6
6
4
0
4
0
5
5
0
0
0
0
0
0
0
0
6
0
0
0
7
0
5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
5
0
0
0
0
0
6
0
0
0
6
0
0
0
0
0
0
0
6
0
6
0
0
0
0
0
6
0
8
0
8
0
8
5
0
0
0
0
6
0
0
7
5
0
7
0
7
0
7
7
0
0
0
0
0
0
6
0
6
0
4
4
0
6
4
2
0
0
0
0
6
2
0
6
6
0
2
0
6
6
2
6
0
0
6
0
0
6
6
6
0
0
6
4
8
6
6
8
8
6
8
6 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0 0
0
0
0
0
0
0
1
0
0
1
0
0
0
0
0
0
Unattended interface:
(220 instances) design
0 0 0 4 5 0 0 4 3 0 0 0 0 0 0 0 7 0 0 0 0 0 0 6 0 0 0 0 0 4 5 6 6 0 0 6 4 7 7 0 0 0 0 0 1 1 0 0 1 0 1 1 1 1 0
0 0 0 0 0 0 0 0 0 0 0 0 6 4 0 0 4 0 0 0 0 0 0 3 0 0 0 0 0 0 0 6 0 0 0 6 0 2 6 0 0 0 0 0 0 0 0 0 1 1 0 1 1 0 0 0 0 0 0 0 0 0 0
4 2 0 6 6 0 0 6 2 0 0 0 8 8 0 0 0 0 7 0 7 4 0 7 0 0 0 7 9 4 5 6 6 6 5 0 8 8 8 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 6 0 0 0 2 2 6 0 0 0 0 0 0 0 0 4 0 0 0 0 0 4 0 6 0 0 9 0 4 0 0 0 5 7 7 0 6 4 4 0 6 0 0 6 0 0 0 0 0 0 1 0 0 0 1 0 1 0 1 1 1 0 0 0 0
0 0 0 0 0 0 0 0 0 4 3 0 0 8 6 0 0 0 0 4 0 0 4 0 0 0 0 0 0 0 0 0 6 6 0 0 0 0 8 6 6 6 6 0 0 0 0 4 1 1 0 0 0 0 1 0 0 0 1 0 1 0 1 1 0 0 0 0 0
4 2 0 0 0 0 0 0 0 5 3 0 0 4 0 0 0 0 2 0 0 0 2 0 0 0 0 8 0 8 8 0 3 0 0 0 6 0 8 4 6 6 6 4 6 6 4 0 0 0 0 0 0 0 0 0 0 0 1
2
0
0 0 1
0
0
0
1
0
1
0
0 0 0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
interface identified by system
architects that lacks corre-
0 0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
1 1 0 0 0 0 1 1 0 0 0 1 0 1 1 0 1 0 0 0 0 0 0 0
1 0
1
0
0
0
1
0
1 0
0 1 1
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0 0
0
0
0
0
sponding team interaction
Team Interaction Matrix 0 0 1 0 0
0
0
0
1
0
0
0
0
0
0
0 0
0
0
1
1
0
0 0
0
0
0
0
0
0
0
0 0 0
0
0 1 1
0 0 1 0 0 1
0
0 1 0
0
0
0 1
0
0 1 1 1
0
0
0
1
1
0
0
0
0
0
0
0 0 0
0
0
Unidentified interface:
0 1 1
0
1
0
1
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
1
1
1
0
1
0
1
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
1
0
0
0
0
0
0
1
0
0
0
1
1
0
0
0
0
0
0
(74 instances) team interac-
1 1 1 1 1 1 0 1 1 1 0 0 1 0 0 0 0 0 1 1
1
1
1
1
1 1 1 1
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
0
1 0
0
0
0
0
0
0
0
0
0
1 1
tion that takes place without a
corresponding design interface
0 0 0 0 0 0 0 0 0 0 0 0 0 1
1 1 1 1 1 1 0 0 0 0 0
1 0 0 0 0 0 0 0 3 0 0
1 1 1 1 0 1 0 0 0 0 0 0
1 0 0 0 0
0
0
Lack of interdependence:
(2,219 instances) components
that do not share an interface or
0
0
Design teams reported the involve design team interactions
occurrence of 423 technical
0
For details, refer to M.E. Sosa, S.D. Eppinger, and C.M. Rowles, “The Misalignment of Product Architecture and Organizational Structure in Complex Product Development,”
Management Science 50, no. 12 (2004): 1674–1689.
tion channels (teams sometimes pass compressor, for example, were virtu- involved teams from different parts
technical information through other ally unchanged from a previous en- of the organization that had fewer
intermediary teams rather than inter- gine generation. As a result, some of opportunities for informal communi-
act directly), the presence of interface the teams responsible for these com- cations, which might have uncovered
carryover (interfaces that have been de- ponents had to pay only marginal at- changes to the previous standards. The
fined in previous projects may not need tention to coordinate their interfaces impact of these unattended interfaces
to be reconfirmed in designing the new among themselves, though they still probably resulted in very small reduc-
project), and the use of interface stan- needed to coordinate with other highly tions in performance or durability of
dardization (interfaces whose specifica- redesigned components. Such inatten- affected components and systems. But
tions have been formally documented tion trickled over into their communi- given the 25- to 30-year life expectancy
are supposed to remain unchanged). cations involving other highly designed of a product like the PW4098, even
In the case of the Pratt &Whitney components. In addition, many struc- small performance deviations could
project, some mismatches occurred be- tural and thermal design interfaces add up and cause significant warranty
cause component designs were carried were left unattended because they or service expense over the life of the
over from previous designs. A number were noncritical or assumed to be stan- product. For example, if a critical com-
of components of the high-pressure dard. In many cases, these interfaces ponent like a turbine airfoil misses its
then decide how to deal with them. other components. There were fewer to
www.HBSPress.org
be missed, and the workload associated properly attended to. In the PW4098 to complete work, the various tools and
with fewer direct and indirect inter- project, the secondary airflow team platforms must be carefully integrated.
faces was more predictable. Too much (one of the six integration teams) was In planning the development of the en-
modularity, though, can lead to myopia, responsible for managing the engine’s gine project that followed the PW4098,
particularly at the subcomponent level. multiple internal thermal and pres- for example, Pratt & Whitney linked full
At Pratt & Whitney we found that the sure management systems to optimize engine aerodynamics and secondary air
design teams of highly modularized engine aerodynamic performance and flow analytical models with component
subsystems were less likely to talk to component durability. It regularly set models to help the design teams man-
teams working on other modularized up meetings and other communica- age their interfaces with the support of
subsystems than were teams working tions between otherwise unconnected the integration teams. Specific people
on more integrated subsystems. In go- teams to address critical interfaces. on each team were then tasked with
ing modular, therefore, product de- Unfortunately, most integration tracking the impact of design changes
signers still need to pay careful atten- teams focus on milestone planning across the component and system mod-
tion to critical interfaces across those and resource allocation, paying only els and communicating those findings
subsystems. marginal attention to the quality of to their counterparts on other teams.
2. Form teams to handle misman- communication between component •••
aged interfaces. Managers also have the teams. That needs to change. Consider Our approach provides a systematic
option to manage critical interfaces – the pain that could have been avoided method for an organization to learn
to ensure that unidentified ones occur in the last phases of the development how and where it is exposed to the risk
or that unattended ones are formal- of the Airbus A380 if one of the integra- of communication failures between
ized – by assigning such work to the tion teams had realized that the elec- design teams working together to de-
teams already tasked with the inter- trical harnesses team in Germany and velop complex products. Moreover, an
action or by making people on the in-
volved teams formally and actively ac-
countable for the interfaces. We would Many design teams miss interfaces because project
recommend this approach for manag- planners haven’t thought through their use of
ing identified interfaces across bound- communication tools and shared platforms.
aries – the interfaces design teams are
most likely to ignore.
Another way to handle missed inter- its counterpart team in France, which organization can use our tools to deter-
faces – one that also avoids the need to were responsible for different sections mine how changes in system architec-
significantly change the organizational of the fuselage, were not properly com- ture, or the emergence and removal of
structure – is to extend the responsibil- municating about their design interface interfaces between system components,
ity of existing integration teams. Most specifications. will affect its ability to avoid communi-
large projects have such teams: At Pratt 3. Select appropriate communica- cation failures in the future. By using
& Whitney there were six teams man- tion support tools. Many design teams DSMs to document the architecture of
aging system issues like air and fuel ef- miss interfaces because project planners the product for every generation of a
ficiency, which affected the design of haven’t thought through their use of product family, managers can identify
practically all engine components. Even communication tools and shared plat- key differences between old and new ar-
though the management of team com- forms. At Airbus, one of the main rea- chitectures. With the alignment matrix,
munications is not usually the primary sons for the communication breakdown managers create a compact and visual
function of integration teams, by the between the A380 teams was the lack representation of interfaces and interac-
nature of their work, these teams com- of compatibility between the computer- tions that allows them to diagnose how
municate with almost all other teams aided design (CAD) tools they used. their organization addresses design in-
in the organization. Accordingly, they Being smarter about using communi- terfaces. Most important, the alignment
are in a position to learn in real time cation tools doesn’t have to involve a lot matrix can help managers properly di-
about the status of critical interfaces of technology: Pratt & Whitney requires rect their efforts to align team interac-
during the design process and to bring teams to regularly complete controlled tions with design interfaces to prevent
unconnected teams together to handle interface documents and component costly problems from occurring later in
critical interfaces that need special at- requirements documents (specifica- the product life cycle.
tention. These integration teams could tions) to ensure that critical interfaces
be made responsible for flagging those are identified and attended to. In cases Reprint R0711J
critical interfaces that are not being where team members use technology To order, see page 155.