Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Improving Work Process Information Flows:

A Case Study
Rebecca Wheeler Melody Chang Susan Volk
Systems and Software Division Enterprise Engineer Division Enterprise Engineer Division
Jet Propulsion Laboratory Jet Propulsion Laboratory Jet Propulsion Laboratory
Pasadena, CA USA Pasadena, CA USA Pasadena, CA USA
Rebecca.wheeler@jpl.nasa.gov Melody.Chang@jpl.nasa.gov Susan.Volk@jpl.nasa.gov

Abstract - The lessons learned from a grass roots effort simultaneous meetings between different members of the
involving information systems developers and engineers rapid design team and their multiple customer teams, to
in the design and development of a customized work track information gathered before and generated during a
process support system, named Fredrik, are discussed. real time design session, and finally to facilitate the
The information system developed is a dynamic, web- storage of all related design documentation and metadata
based work process system to support the end-to-end so it could be subsequently accessed by the design team
scheduling, planning, design and documentation activities and their customers for the study.
of an “extreme collaboration”design team. The system is
used to help schedule and manage multiple pre-design Fredrik was developed for use by Team X. Team X
session meetings, to track and store information gathered is a rapid design team utilized at the Jet Propulsion
and generated before, during and after a real time design Laboratory (JPL) to rapidly complete designs in support
session, and to facilitate document and design of space mission proposals. Team X employs an
information retrieval for later re-use. While the close “extreme collaboration” design paradigm (1), which
partnering between the users and developers solved some involves the co-location of a team of about 20
long standing trust issues, and resulted in a system more multidisciplinary engineers, and their customers, for a
optimized to the engineering team’s needs, the short period of time in a technologically intensive
participatory design process inadvertenly produced a physical environment to complete a mission design in real
less stable and sexy system potentially affecting future time.
sources of funding for upgrading the system.
2 Team X Work Process Structure
Team X achieves “extreme collaboration” during
Keywords: concurrent engineering, rapid prototyping, three half-day concurrent design sessions, that take place
user centered design, real time design, work process over the course of a week. During these sessions, all of
support systems, case study. the team members are co-located in a design center that
contains networked workstations and the supporting data
management infrastructure, large interactive graphic
displays, computer modeling and simulation tools, and a
1 Introduction mature shared generic project model that the design team
What happened when a well established rapid populates in real time. The team members are fully
advanced design team with highly dynamic work committed to completing the design work during the
processes and inadequate data collection practices was concurrent design period (2). While the concurrent
partnered with a rapid prototyping software design team? design activity is the most innovative part of the Team X
A heavily utilized customized scheduling and tracking design process, a typical design actually requires at least
system tightly locked to the rapid design team process three weeks from start to finish. Pre design session
was the result. This paper presents the lessons learned activities include design session scoping, planning and
from involving the users in the design and development pre analysis. Post design session activities include design
of a customized work process support system named analysis wrap up and documentation. The pre and post
Fredrik. session activities typically take place “offline” using more
traditional asynchronous team product development
The basic requirement satisfied by Fredrik was to methods.
collect, organize and retrieve information generated
during the life cycle of a rapid design team process. The Most Team X members simultaneously work on
system is used to help manage the status of multiple multiple studies. During any given week, a given team
member’s activities could include pre-session planning harder to find and re-use data than to "reinvent" the data.
and design meetings for multiple studies, participation in The rapid prototype software design team was asked to
concurrent design sessions, and completion of several assess the current Team X processes and recommend an
study reports from previously completed design sessions. information system approach that would address these
Meanwhile, new and existing customers continue to "hidden costs".
request new design sessions or changes to their already
scheduled planning and design session schedules. With The software design team performed a search and
the exception of the Team lead and an administrator, none evaluation of COTS scheduling and data management
of the team members work full time as Team X members. software. There were many products available that
Furthermore, their work load involves almost daily addressed scheduling but lacked the data management
changes in the studies that they might participate in for as capabilities needed by Team X. Conversely, there were
few as three weeks. Therefore a big concern is how to many products available that addressed data management
keep the Team and the customers “on the same page” in but lacked the scheduling capabilities needed. The option
such a dynamic environment. of integrating and/or modifying COTS software was
considered. However, the cost savings associated with
3 Fredrik: A Work Process Support the use of COTS software can be lost if the software is
not implemented "as is", the software was not designed
System for Team X for "re-use" or more than 45% of the software needs to be
From the perspective of time and cost reduction, modified (4). Also, the Team X functional requirements
Team X has been extremely successful. Well over 600 for scheduling and data management were not well
early-phase project designs have been completed over the defined at the time.
course of nearly ten years. These studies were completed
in less than one-tenth the time of the previous process at Given the lack of a "perfect match" COTS tool and
JPL, and for less than one third of the cost (3). However, the vague system requirements, the software team
prior to the development of Fredrik, Team X had no decided to develop a system through rapid prototyping
systematic information system to manage its work flow. for the Team X scheduling and metadata management
Scheduling was ad hoc—handled through phone calls, capabilities and use the JPL institutional COTS software
emails and multiple, internally inconsistent, methods of for the file management. By using rapid prototyping, the
posting schedules. All the information generated from software design team could interact with Team X to
meetings with the customer, from customer presentations, provide an information system quickly, as well as help
from the team design parameters and from product with defining the Team X functional requirements. Thus
iterations were stored under one customer file maintained Fredrik was conceived
on a server. When the server became “full”, the files
were moved to compact disks (CDs). If the customer Fredrik is a dynamic, web-based work process system
wished to retrieve a copy of a previously completed to support the Team X end-to-end design process. Its
report or the Team wished to “re-use” previously essential elements are a series of standardized on-line data
completed study data, a team member would have to sort collection forms integrated with a relational database that
through some 300 CDs that contained the study results. supports detailed scheduling, group lists, automated email
Further, with no metadata or systematic file storage and notification, collection of design status and standardized
close-out procedures, she would often have to “guess” metadata information, as well as a commercial document
which were the final products and assess whether stored storage system. Fredrik enables tracking of a study
data products had been made internally consistent. Thus, design from the time a customer requests a study until the
while Team X was meeting the original goal of rapid time the final report is stored in an on-line library, along
design at greatly lowered costs, some of the “hidden with all the metadata collected over the “lifetime” of the
costs” of that process as implemented were study process. While the scheduler can be viewed by
inconsistencies in the scheduling process, customer data anyone with access to internal networks, users must
collection process, team briefing process, and product otherwise log in via a secure web interface to access any
production, storage and retrieval processes. These of the other information in the system. Users are
inconsistencies were an “expense” imposed on team authenticated and provided role based authorization to
members, who spent their limited design time retrieving access different resources of the system.
information from previous study sessions that lacked
adequate metadata which should have been added at the From Team X’s perspective, the heart of the Fredrik
time the information was generated. system is an on-line calendar/scheduler that provides
scheduling and displays all scheduled pre-sessions,
Team X leaders recognized that the "hidden costs" design sessions, product due dates, and personnel
of inadequate information management was growing as availability (Figure 1). Clicking on any particular design
the number of studies and customers increased. It was session from the calendar brings up scheduling event and
personnel assignment forms (Figure 2), customer study study close-out checklist form (Figure 4).
request form (Figure 3) documentation status forms and

Figure 1: Team X Calendar

Figure 2: Scheduling Event and Personnel Assignment


Form

Figure 3: Customer Study Request Form


Figure 4: Study Status Checklist

While the schedule can be viewed by anyone with session dates (Figure 5). The email notification thus acts
access to the JPL network, only authorized Team as a trigger during the process flow, facilitating
members are allowed to schedule sessions and other communication between Team X members and their
events by logging in via a secure web interface. customer. Any schedule changes can result in additional
emails which, though not automatically sent, can be
When a session is scheduled, an email is easily initiated by the push of a button by the
automatically sent to affected team members and the administrator.
customer notifying them of the pre-session and study

Figure 5: Email Notification Form


All the data that customers enter, including the product which is located in a secure commercial on-line
email log information, is also stored in a relational library system that accommodates role based access to
database. This not only helps the Team X customers different studies. All of the studies generated in the last
keep track of the study status, but it also helps the Team nine years and their associated metadata have been added
X members to retrieve and re-use the data. All metadata to the commercial document management system so that
collected over the lifetime of the study utilizing the they can be instantly retrieved by the customer (if
tracking forms is stored along with a link to the final authorized) or by team members for re-use (Figure 6).
Figure 6: Studies and Study Progress List

4 Lessons Learned than necessary to do the job they need to get immediately
done, no matter what the greater “need” articulated by the
Fredrik was developed over several years using a developers or outside managers. Finally, Team X has
participatory design and rapid prototyping methodology. endured a long history of development efforts that had
Significant improvements were gained in process tried to impose their “improved” process or technology
consistency, efficiency, and information retrievability, but solutions on the Team without any understanding of their
the development process required a lot of interaction constraints, processes or needs. As a matter of course,
between the Team leaders and the information system therefore, Team members understandably resist changes
developers to articulate and automate the study session that they perceive might inhibit their effectiveness, even
process. However, without that interaction, the if the change might resolve issues they really want
information system would not have been optimized to resolved.
resolve Team X’s most pressing problems. Indeed, the
Team had already rejected the adoption of tool several The solution to this legitimate resistance was to
other efficiency or design support tools developed establish regular interactive sessions between the Team X
without its participation. Today, Team X consistently leaders and the Fredrik developers. The Team X process
uses Fredrik to manage all of its scheduling—pre- was captured in a way that could be implemented in
sessions, team assignments, design sessions and product Fredrik. The developers implemented practical solutions
production status. to some long-standing Team X problems. This process
established trust between the two groups.
The participatory design process required constant
effort on the part of management to maintain because Rapid Development Prototypers MUST anticipate
there was often tension between the development and and accommodate entire database redesigns.
engineering teams’ views of needed functionality. In
general, the Fredrik team wanted to “push the envelope” The good news was that Team X leadership really
in terms of process redesign and information system wanted support and were willing to standardize and even
capability. Team X, generally wanted to simply automate evolve their processes to be supportable by enhancements
their existing process. Management’s job was to try to in Fredrik when the enhancements really solved their
manage the tension and amicability between the two problems. The bad news was that with the new sense of
teams in order to effect the best compromise. Three “trust” between the teams, the Team X leadership started
major lessons learned from implementing this relatively requesting more and more design changes. Indeed, more
uncomplicated system are discussed below. than a year was spent simply adding “new functionality”
to Fredrik. While the development team was delighted
Information system users and information system that their product was being adopted by Team X, they
developers MUST continuously interact during the were becoming frustrated that Team X was specifying
information system development process. “solutions” that were essentially work-arounds to an
increasingly unwieldy prototype database structure.
Team X has work to do. They don’t have time to Instead, the developers requested down time to redesign
play with what they perceive as “esoteric” tools that don’t the databases. However, the Team X management team
seem to solve their problems or require a lot of effort to was used to getting at least some of their new
learn. Further, they didn’t have time to “populate” ANY requirements implemented every few weeks and resisted
data collection information systems with far more data freezing the current system for a redesign.
Coincidentally, a Team X reorganization occurred, and intensively utilized by them. However the unanticipated
the new Team X managers were responsive to consequences were a less stable and “sexy” system than
information system developer needs and the existing might have been developed in the absence of a user based
Fredrik configuration was “frozen” for a few months design. One fallout was difficulties in finding funding
while the redesigned database are restructured. It remains sources for early development of the system and,
to be seen, however, whether the original camaraderie potentially affecting future sources of funding for
between the mission design team and the information upgrading the system.
system design team will be re-established with the new
Team X. Currently the Fredrik database system is being
modified to accomodate the addition of new rapid design
Funding sponsors NEED to recognize that “sexy” teams and management structures, which are likely to be
technologies are not necessarily what is needed to more réceptive to the adoption and funding of Fredrik
resolve process inefficiencies. than was the case in earlier years. It remains to be seen,
whether the original camaraderie between the new
Obtaining, retaining and getting enough funding for mission design teams and the Fredrik design team can be
the development of Fredrik—especially before re-established.
demonstrating usefulness to Team X members—was a
constant struggle. The system is not particularly exciting
from a technology development perspective—no artificial References
intelligence, no optimization algorithms—just forms,
database, a scheduler and code to link all the pieces
together. Therefore, winning external or internal [1] Mark, G. (2002). Extreme Collaboration.
technology development proposals during the initial Communications of the ACM. Vol. 45(6), pp.89-93.
development phases for Fredrik were frequently
unsuccessful. Institutional funding sources favored the [2] Chachere, J., Kunz, J., and Levitt, R. 2004.1, “Can
development of better engineering tools, focusing on You Accelerate Your Project Using Extreme
further improving concurrent design session Collaboration? A Model-Based Analysis” 2004
effectiveness, so those funding sources were frequently International Symposium on Collaborative Technologies
unsuccessful as well. The hidden costs to the design team and Systems; Also Center for Integrated Facility
members inherent in their process inefficiencies was Engineering Technical Report T152, Stanford University,
evident only to a few of Team X customers, technology Stanford, CA
developers and, finally, concurrent Team managers, who
after continued lobbying on the part of the original [3] Wall, S., Reinventing the Design Process: Teams
Fredrik design team, finally banded together and and Models," International Astronautical Federation
provided enough seed money to implement the system. Specialist Symposium on Novel Concepts for Smaller,
However, without a strong belief in the value of a work Faster and Better Space Missions, Redondo Beach, CA,
process support system on the part of the original Fredrik April 1999.
design team, Team X could still be manually scheduling
its meetings and sorting through hundreds of CDs looking [4] Boehmn, B. Abts, C., Brown, W., Chulani, Sul,
for its data. Clerk, B.K., Horowitz, E., Madachy, R., Reifer, D.J. and
Steece, B., Software Cost Estimation with COCOMO II,
Prentice Hall, 2000.
5 Conclusions
Acknowledgements
Not long ago, Team X had to call or email people
to schedule their meetings and hunt through hundreds of The research described in this publication was carried out
CDs for previous reports. Even though their process was at the Jet Propulsion Laboratory, California Institute of
very inefficient, they had experienced many problems Technology, under a contract with the National
with previous improvement attempts and hence resisted Aeronautics and Space Administration.
using new tools. A close partnering with the Fredrik
developers solved some practical problems and built trust
allowing further development to address inefficiencies.

The rapid proptotyping approach was ideally suited to


articulating Team X requirements and assuring that they
were implemented in a manner useful to Team X. This in
turn resulted in an process management system that was

You might also like