This article originally appeared
in the August 1999 issue of PDBPR
High-Velocity Development:
Hewlett-Packard Veterans Share Lessons
From Supply Chain Management
Scott Elliott and Brian Hughes, who jointly have more than forty years of
experience with Palo, Alto, CA-based Hewlett Packard, think that when it
comes to shortening product development time there is a lot to be learned
from supply chain management. Elliott, who holds a doctorate in electrical
engineering, held senior management positions in R& D and manufacturing at HP for more
than two decades before recently joining Boston, MA-based Product Development Consulting.
Hughes, who also holds a Ph.D., is Engineering Productivity Manager with Agilent (formerly
HP's Test and Measurement Organization). His team is chartered to enhance the design
processes for faster time-to-market and higher engineering efficiency for 400 engineers.
BPR: What's your take on
supply chain management?
SE: "Many R&D managers
try to reduce their product development time and expense. They believe the path to shorter
development projects is to freeze a full set of product specifications as early as
possible, then maximize resource utilization over all parts of the project at once. This
form of 'push' management can slow project execution times radically.
"Development projects are much more rapid
and more successful if they follow a few lessons from supply chain management: move risk
as far upstream as possible, and move customization decisions as far downstream as
possible. Supply chain management is the business of optimizing physical inventories and
supply logistics for maximum customer service levels at a minimum cost. Supply chain
managers strive for high product 'velocity', rapid movement of inventory through the value
chain."
BPR: What connection do you
see between supply chain management and product development flow?
BH: "The purpose of a
development project is not to build products or
physical inventories, but instead to build a complete inventory of knowledge
for the manufacturers, suppliers and marketers of the new product. When a
project manager organizes and leads her project as a 'supply chain' for
knowledge, development cost and lead times can be cut and high development 'velocity' is
achieved."
SE: "In the supply chain,
inventories are risky, costing time and money to
build and store. These inventories gain value as they move through the
manufacturing process but, when stored or improperly routed, can lose value
rapidly (or 'rot') due to obsolescence, changes, assembly mistakes,
component flaws, etc. Excess inventory can greatly slow the 'velocity' of
products through the supply chain, bottlenecking capacity, increasing cycle
times and exacerbating the risk.
By analogy, the 'knowledge inventory' of a
development program gains in
value as it flows through the development process, but can also 'rot' if it
is stalled or not properly coordinated with other knowledge streams.
"For example, part of a team might develop
a microprocessor-based controller in advance of other parts of a product. While this set
of hardware and programming knowledge waits for the rest of the project to catch-up, the
microprocessor could become obsolete, or the programmer might leave, or
other factors could cause the controller to have to be redeveloped. Excess
knowledge 'work-in-progress' (too many projects occurring simultaneously)
can also cause bottlenecks in critical development processes and services
and greatly increase the development time and risk of the project."
BH: "Lean manufacturing and lean supply chain techniques have vastly
improved manufacturing techniques in the last decade by minimizing risky
inventory, moving stores as far upstream as practical, and using
postponement techniques for customization. In design, the analogous
practices are to develop the 'riskiest' knowledge inventory first
('upstream'), and to postpone low-risk, customized features and
specifications until later in the project. Knowledge inventory is 'risky' if
the team does not know how long it will take to acquire, or has lower
confidence that it can be adequately acquired."
BPR: Can you give us some
examples?
SE: "At Hewlett Packard's
Microwave Technology Center, I managed a group of engineers that was developing a
state-of-the-art electro-optical modulator
chip for fiber optics communication test equipment. These devices, although
demonstrated in universities and scientific labs, were not in production anywhere at the
time. The modulator is a 'chip' of the crystal lithium niobate a few centimeters on a
side. Light enters one side of the chip though an optical fiber and exits the other side
though another fiber. An electrical connection provides for the modulating signal.
"The development team began to think about
the streams of knowledge
development required to produce a package of documented processes, supplies and skills
needed to market, manufacture, and deliver these chips to the 'customers' instrument
producers. The team first did preliminary studies to assess the highest risk areas for
this project. They found these to be the
optical fiber alignment and attachment method and the reliability of the
fiber connection. Surprisingly, the pattern of the optical waveguide on the
chip was evaluated as much lower risk, even though this "masking" pattern
was the subject of most previous research papers.
"To move these risks upstream as far as
possible, the team began experiments with aligning and attaching fibers to the chips with
very simple
(straight-through) waveguide patterns. As soon as any attachment was made,
it was subject to rigorous environmental and strife testing the team did
not wait to see if modulation could be demonstrated with the masking
patterns. Only after confidence became high from these experiments did the
engineers experiment with more sophisticated waveguide and metalization
patterns to demonstrate optical modulation. This strategy was counter to the
prevailing project management practice, which was to approach the more
'technically stimulating' part of the project first achieving the
high-speed optical modulation.
BPR: What happened?
SE: "After about 12
months, the modulator team had demonstrated enough
confidence and produced enough skills, tools, and information to manufacture
electro-optical modulators. Now it was time to begin lower-risk, custom
parts of the project designing a modulator for a specific application, an
optical network analyzer. The specific waveguide pattern to achieve this
performance was quickly developed and demonstrated by the team, using the
latest university research, while the final specs for the analyzer were
being established. The instrument was introduced close to original schedule
and without problems from the modulator assembly. Other modulators with
customized performance were produced very rapidly and with great confidence for other
customers."
BPR: How does this relate
to your supply chain management analogy?
BH: "A development team
produces knowledge inventory. The knowledge
inventory flows and builds through the development process, increasing
confidence and decreasing risk. Like physical inventory, knowledge inventory
takes time to build and carries cost and risk. Let us illustrate with a
couple of figures. Figure 1a diagrams a 'push' supply chain,
typical of
slower clockspeed industries and vertically integrated companies.
Inventories of parts and work-in-progress are stored in many places so that
all workcenters can be kept fully utilized. Following any one part or
component through the many processes and queues shows that the lead time can be very slow.
"Inventories in such supply chains can be
very risky, rotting rapidly for
myriad reasons. Long test and rework loops can further stretch cycle times.
Figure 1b illustrates a more modern supply chain. Suppliers
provide parts
'just in time' for manufacturing, minimizing inventories and maximizing the 'velocity' of
any part. Final variations in the product may be postponed and made by the
distributor or even by the retailer to minimize inventories of multiple variations.
Components traverse through the system faster than they can become obsolete, and flaws are
discovered rapidly."
"Figure 2a
illustrates a typical 'push' development flow. In the product
definition phase, a big investment is made in R&D and marketing to specify
the product as fully as possible. The concept documents, the reference
specification, and potential technologies and suppliers represent the 'knowledge
inventory' at this point. The development project is usually planned as a parallel
development of all modules of the product.
Subprojects are started as soon as possible.
"Often the individual developers are
working on multiple projects, and key
resources and services are shared with other projects to maximize
utilization. Knowledge inventory is built and "stored" until other
subprojects are ready. The high-risk parts may not be tested until well into
the project or prototype phase, causing long iteration loops. The triangles
indicate storage points for knowledge inventory."
"Figure 2b
illustrates the 'just-in-time' analogy for project management.
First, the knowledge for the concept is collected from marketing, R&D and
management. Next, a careful analysis is done to ascertain the highest-risk
and longest lead-time elements of the project. Most of the project resources
are initially dedicated to these knowledge development tasks, postponing
many of the lower-risk studies. 'Suppliers and subcontractors' of knowledge
may be employed to add important knowledge elements, component information and
documentation to the project where needed. Medium- and low-risk subprojects are postponed
and started as late as possible so that the total knowledge package comes together at the
same time."
BPR: Where does
"postponement manufacturing" enter the picture?
SE: "Postponement
manufacturing is one of the cornerstones of modern supply chain management practice. Many
of today's products offer variations,
options, and customized features. These variations cause difficulties for
supply chain managers, who must anticipate needs for retailers and
distributions centers. It is very expensive to provide all distributors and
retailers with all combinations. Instead, most companies now provide
distribution centers with the basic components. The distributors, the
retailers, or even sometimes the customers do the final assembly. An example is personal
computers, where distributors may assemble a system from choices of microprocessors, hard
disk drives, etc. with very fast turnaround for customers.
"The modulator case study offers an example
of postponement design. The
development team was able to 'stock' an inventory of knowledge about how to
build a reliable modulator, with processing techniques, packaging, and
reliable waveguide attachments before deciding on an actual design spec.
Concurrently, the 'customer' (the instrument development team) was designing an instrument
in need of a particular modulator specification. The modulator team was able to use
already proven CAD techniques to rapidly produce the needed modulator information and to
demonstrate a prototype and production capability."
BPR: What's the problem
with the current approach to project management?
BH: "Most project managers
schedule projects using critical-path analysis
tools with the 'earliest possible' start date in mind. This tendency stems
from deep-rooted fear that all parts of a project may be high risk, so one
should try to find as many problems as possible as early as possible. In
addition, there is a fundamental belief that projects will get done faster
and better if everyone is utilized to the maximum.
"We agree with Don Reinertsen when he
points out--in his book Managing the
Design Factory--that these assumptions cause several effects that reinforce
each other in a spiral of problems and slippage. Lead times and cycle times
are longer because all of the resources and services build queues of work.
Engineer 'shotgun' designs (try multiple variations at once) to try to get
more knowledge out of the slower processes, adding more 'work-in-progress'
to the bottlenecks. The less-risky projects sap resources from the
higher-risk ones, slowing everything down.
SE: "Leading a lean and
rapid development project requires intelligence,
guts, discipline and faith. Intelligence is needed up-front to carefully
analyze the real risks in the project. Guts and discipline are required to
assign most resources to mitigating these risks, almost ignoring the
low-risk bits for the time being. Finally, a project manager must have faith
that the low-risk parts really are that, or at least her lean design team
can solve that unexpected problem rapidly. The risk can be lowered in a
number of ways, including re-use of proven subassemblies, using well-proven
CAD models, etc.
BPR: What's the bottom
line?
BH: "The job of a
development team is to build a 'knowledge' product,
consisting of the complete set of information and skills needed to market,
manufacture, and deliver a physical product. The design chain can be viewed
as a collection of 'knowledge inventory' streams that are building in value
and converging in time to form the final knowledge product. The timing and
flow of these knowledge inventories is very important in minimizing the
development time. Too much "work-in-progress" can cause severe bottlenecks,
slowing the project to a halt.
"The project manager can run a lean
development chain by minimizing project
work and scheduling knowledge inventory to arrive at the right time. The
highest risk investigations should be heavily resourced at the beginning,
with low-risk projects and customization postponed until late in the
process. Just as in running a lean manufacturing line, this kind of project
management requires intelligence, guts, discipline and faith. A case study
from a successful project in Hewlett Packard illustrated these concepts." |
