Lockheed Missile and Space relied on Formtek to provide the platform support for the program
Scott Wallace
For LMSC (Lockheed Missile and Space Corp.), the challenge was clear, though hardly simple. The DoD (Department of Defense) wanted a new missile and wanted it fast. Up for grabs was a $688 million program to not only design and develop a new missile but also demonstrate and validate accelerated engineering techniques that could be used in other DoD programs.
To engineer the missile within the DoD's 24-month time frame--five years is the usual--LMSC needed to change the way it designed and engineered missiles. It also needed a sophisticated engineering data management system, which it didn't have time to develop on its own. The system would have to be able to rou
te engineering documents to engineers working on a variety of platforms to capture their comments and input for the engineering review process. After extensive review of a variety of systems, LMSCfired off a fax to its six short-list vendors in September 1992, which consisted of a lengthy survey of the vendors' systems, software, services, and future strategies. If vendors wanted to remain in the competition, they had to respond within 72 hours. This was LMSC's first test of the vendors' readiness to support a program whose success depends on squeezing time out of an already highly compressed schedule.
Concurrent Engineering
Key to the U.S. Army's Thaad (Theatre High Altitude Area Defense) "kinetic kill" interceptor missile program is concurrent engineering or, as it is called at LMSC, IPD (integrated product development). An Institute for Defense Analysis report, The Role of Concurrent Engineering in Weapons Systems Acquisition, defines IPD as "a systematic approach to the integrated, concurrent de
sign of products and their related processes, including manufacturing and support. This approach is intended to cause the developer, from the outset, to consider all elements of the product life cycle from conception through disposal, including quality, cost, schedule, and user requirements."
The Thaad program is LMSC's first production use of IPD. Since the award of the Thaad contract to LMSC in October 1992, the program has depended heavily on a combination of methodology and technology to stay abreast of a 24-month schedule that calls for design, development, and prototype manufacturing of a completely new type of missile (see the figure "IPD Uses Teaming and Technology to Compress Time"). In the past, this phase would have had a duration of five years or more.
LMSC relied on its IBM 3090-based computer-integrated engineering and manufacturing systems during previous missile development programs and initial preaward planning targeted them to support the Thaad project as well. Problems with th
is approach surfaced quickly. One problem was cost: Mainframe services turned out to be beyond the budget of the program, which targeted client/server price and performance and was built on client/server technology. Another more important problem was functionality: The mainframe system was geared toward local workgroups and could not be suitably restructured to support the large, geographically dispersed teams on this highly classified project. In addition, the Thaad project called for multicorporate teaming--something new to LMSC on this kind of program. Efforts to modify and enhance in-house mainframe services to support the Thaad development agenda were unsuccessful. However, in the process, a clear understanding of the program's needs was developed, making LMSC well prepared to specify its system and evaluate vendors to support the Thaad project.
Thaad Information Management System
The principal TIMS (Thaad Information Management System) selection criteria included seamless support of existing a
pplications and a uniform user interface--no matter which of many Thaad platforms the application was running on. Within a month of sending out its fax survey, LMSC had made a decision. From a list of vendors that included IBM, Control Data, SDRC (Structural Dynamics Research Corp.), Sherpa, and Interleaf, LMSC selected Formtek, a Pittsburgh-based engineering document management vendor that spun-off from Carnegie Mellon University in 1982. The approximately $2.5 million contract called for Formtek to provide LMSC with software (40 percent of contract value), as well as software development and support services (60 percent of contract value) over a two-year period. No hardware was involved in the deal.
To keep this complex project as simple as possible, LMSC basically layered Formtek's engineering data management software onto its own networks and desktop infrastructures--which LMSC was expanding to support the Thaad project. The resulting system would support Mac, Silicon Graphics, DEC, Hewlett-Packard
, IBM RS/6000, and PC workstations, as well as all the information environments and data types found on the Thaad project. Users primarily run engineering applications supported by FDDI (Fiber Distributed Data Interface) backbones for data transfer and Ethernet LANs for user connectivity. Sited on LMSC's campus in Sunnyvale, California, the Thaad program is classified and operates out of restricted-access facilities replete with "secure and reliable" networks and computing systems, in addition to a "data fire wall" that protects the project's information resources from intrusion.
You Don't Have to Be a Rocket Scientist
The principal purpose of TIMS is to improve and expedite the design-review process. This boils down to managing a wide variety of electronic documents and supporting their on-line review on various platforms by many different people, who can't or won't invest the time to come up to speed on applications. "These people are rocket scientists. They're busy, and they don't want to learn h
ow to use some new piece of software," said Mark Kadrich, network systems manager for the Thaad program at LMSC. "Some users are just beginning to master E-mail, messaging, and conferencing. Layer on top of that an automated meeting system, and you have some very confused users."
The combination of novel applications, mixed user skills, and an extremely tight development schedule helped to keep everyone focused, including those in charge of developing and supporting TIMS. "The goal was to help the program: We were not trying to develop an information management system for its own sake," noted Kadrich. Building the review environment on in-place networks, workstations, and user-applications skills meant that a minimum of implementation support and training was required--and a minimum of user resistance was generated. All this was critical, because the system needed to be operational and pulling its weight less than four months from the contract award.
Partnering for Real
Another important
selection criterion was the vendor's willingness and ability to support the needs of the Thaad development effort--tested many times during the procurement process. "At the beginning of the program, not all of the technology was available to do what we needed," Kadrich said. While Formtek's software supported a design-review process, its support wasn't optimized for the IPD processes at LMSC or for the Thaad-style review the project required.
That wasn't the only vendor issue on this fast-moving, highly adaptive program. "For us, requirements are going to change; we honestly believe that the implementation should be iterative. We knew that as program requirements changed and user requirements changed and the users began to really understand the system, we were going to have to go back and retool," said Kadrich. User acceptance--from a community ranging from computer illiterates to computer masters--was critical. "We felt that by getting the users to buy in from the very beginning--finding out what the
y thought was important, changing the system, and feeding it back to them--we would be more successful."
How could LMSC go about getting the vendor support it needed when it was needed, if the nature of that support kept changing? "We decided to make the vendor part of the team," explained Kadrich. "We really had to work together in an environment that includes just about every discipline you can imagine. So we sat down and from scratch, developed a module that allows people to check in, review, mark up, and circulate a document in an integrated product development environment."
To do this, LMSC selected one of the program's four product development teams to help. "We chose the interstage product development team as a model," said Barbara Victorino, Thaad system administrator. The interstage is a complex portion of the missile and would present broad challenges to the review effort. Victorino adds, "We went to them and asked what would help them do their job better and faster; what would help th
em communicate more efficiently with each other?" Based on these iterative discussions, the design-review module specification was completed.
Formtek played an active support role during the design-review module requirements analysis, but followed LMSC's lead when it came to specifying the system. "The Thaad program provided a solid opportunity for us to make enhancements in our system--enhancements driven directly by customer requirements," said James A. Bole, vice president of engineering at Formtek. During this development effort, Formtek extended its platform support and enhanced its user interface--efforts designed to better support Thaad design-review requirements.
Design Review: Old and New
In development efforts preceding Thaad, a set of engineering drawings would be posted on a sign-off board and engineers from all disciplines would review it and mark it up. Over a two- or three-week period, designers representing six disciplines would "redline" the drawings with their comments
and corrections. Then, the principal designer would incorporate those changes into the original design, post the drawings on the sign-off board, and the process would start again. When a drawing set was reviewed without change, it was accepted. Typical duration for a design review was two weeks.
Under the Thaad program, a new review regimen was instituted. (And a good thing, too, because the program has completed more than 4000 design reviews since October 1992.) Because the TIMS design-review system was built around workstations, networks, and electronic documents, it supported on-line review, making it possible to route the drawings to the people, rather than the other way around. Furthermore, reviewers could request and review design documentation from their own terminals when they wanted. The net result, according to Kadrich and his staff, was a dramatic shortening of review duration (to as little as three hours in some cases) and a dramatic improvement in review--and design--quality.
The TI
MS design-review process has three basic steps: review, comment, and vote. Design reviews can be formal or informal. If the design review is formal, it includes a binding "vote," during which each reviewer registers his or her judgment as to whether the design is "flight ready" or not. If the review is an informal gathering of opinion and comment, voting is not required.
At the beginning of a review, users receive an E-mail alert that materials are staged and awaiting review. On request, reviewers are presented with raster images of documents on their workstations and use redline tools to mark up the documentation--from J-size engineering drawings to standard-size pages of text.
During the comment phase, users enter their comments concerning the drawings, specifications, and so on, for inclusion in the formal record of the review. This repository of comment information turns out to be very valuable for two reasons. First, it enables the historical capture of information that is driving the desig
n, rather than capturing just the end-states of the review process. This makes it easier to identify improvements--and problems--and trace them back to their origins. Second, it enables LMSC to enhance and improve the Thaad review process itself. "One of the challenges we have is to develop a better way to do business," said Victorino. "TIMS is helping us take each step of the process, analyze it, and gather data." The result is ongoing improvement in the design-review process.
Wisdom Archive
Maintaining a complete and comprehensive log of review, comment, and vote activities does more than simply "record" the project, it enables the project to be "played back in reverse." This has some interesting benefits. "We went through a design cycle a few months ago that appeared to be a dead end: It looked like we had to go back to square one," Kadrich said. Before reengineering a major portion of the missile system, staff members "replayed" the engineering process, reviewing the considerations and deci
sions that had been made along the way. This convinced LMSC personnel that they were, in fact, on the right track and the design was resubmitted to the customer. "They were more comfortable with the idea this time round. Rather than weeks and weeks of redevelopment, it took a couple of hours to reload the data files on the server."
For project managers, design-review status reporting is like a finger on the day-to-day pulse of the program: It gives them a detailed understanding of the current state of the project and creates the opportunity to take corrective action early, if necessary. Accurate, up-to-date project status is particularly critical to informed decision making on a crash program like Thaad, where a task slip of two weeks would prove disastrous to the entire project.
Finally, by archiving the review process and the comments registered during it, the TIMS captures for perpetuity a substantive and previously short-lived knowledge base. The design-review module stores information conce
rning why certain changes were made--or not made--so that, should team members retire or otherwise be unable to help reconstruct the logic of certain decisions, the knowledge base is there, and the process that transpired to result in that knowledge can be reviewed.
Because of the IPD structure of the project, this knowledge base has a great deal of applicability (see the figure "IPD Technology"). "A given team is responsible from concept to design to flight," noted Victorino. "[It] has the authority to make design, schedule, and budget commitments. This is very different from the way we did business in the past." Because of the novel structure of the program and the new E-mail format of project communication, tracking these details is critical both to ensure schedule and budget commitments are met and to understand how they were met.
Platform Issues
There are nearly 800 LMSC employees involved in the Thaad program, slightly more than a third of whom TIMS supports locally. The bulk of th
e user workstations are Macs running System 7, but Unix and Windows are also supported. The most common TIMS platform is a Mac IIci. The program uses over 220 Macs, typically with 8 MB of memory, an 80-MB hard drive, and either a 13-inch or 16-inch monitor (some are monochrome, some color).
Another popular platform is the SGI Indigo II Elan--typically configured with 64 MB of memory, 1.3-GB hard disks, and 19-inch color monitors; SGI platforms account for over 30 workstations. There are eight HP workstations dedicated to the Thaad project, typically model 750 systems with 128 MB of memory, 2.4-GB hard disks, and 19-inch color displays. Six IBM RS/6000 platforms and a mix of PCs round out the workstation population that is dedicated to the Thaad project. Other, more casual users have a similar mix of platforms.
Initially deployed only on the secure side of the information fire wall, TIMS will soon support access to the unclassified side as well. This is to be accomplished by a security system tha
t safely penetrates the fire wall. Today, a prototype connection has been tested in one direction: Data can move from the unclassified side to the classified side only. A secure, bidirectional channel is in the analysis and planning stage. "We use a Motorola Network Encryption System that has been approved by the National Security Administration for processing information classified all the way up to 'top secret' level," said Kadrich. A Sun Microsystems NFS link supports secure connectivity to networks on both sides of the fire wall.
A pair of HP 9000/827 servers (both have 128 MB of memory; one has two 1.3-GB disks and one has three) support users, one in the classified and one in the unclassified domain. Two FDDI backbones are connected to two sets of four Ethernet LANs by hot-redundant routers. Because engineering drawings of 8 or 10 MB are common, keeping this weighty traffic on the FDDI backbone and off the Ethernets improves the response of the user-network segments.
Originally, TIMS inclu
ded an optical jukebox for bulk storage. But users found the 15- to 30-second response time to be unworkable, and the jukebox itself proved to be the least reliable piece of equipment in a program that required better than 99 percent uptime. As a result, the optical jukebox was scrapped in favor of additional magnetic storage, most of which was installed at the server level.
Of the on-line TIMS data, about 20 percent is text: contract documentation, design commentary, product or process specifications, or other word processed materials stored in both native and PostScript formats. Approximately 75 percent of the TIMS data is engineering drawings saved in raster format. The remaining 5 percent of on-line data consists of scanned images derived from hard copy that Thaad contractors--who are unable or unwilling to electronically deliver drawings and support materials--provide.
Data-conversion routines are critical to supporting the wide variety of platforms and applications within the Thaad program
. While engineering drawings are developed and archived in their native application formats, drawings are distributed (for review and commentary) in their raster versions, ensuring they can be viewed from a combination of platforms. Conversion between file formats, such as ASCII, CCITT Group IV, CALS (Computer-Aided Acquisition and Logistics Support) type I, HPGL (Hewlett-Packard Graphics Language), and PostScript, enable document interchange and flexibility at the application level.
Document scanning takes place on a scanning subsystem. A pair of NEC 386/33 machines are scan servers for a large-format scanner from Vidar and an A/B size scanner from Fujitsu. A Sun Sparcstation 2 is used to review scanned images, to make any corrections or edits to the images, and to index and commit the images to the TIMS database.
As might be expected, printing Thaad documents comes with its own overhead and headaches. "Classified document generation brings up a lot of problems," confirmed Kadrich. "We control
printing through a central location." Documents are output on a QMS printer (for A/B size documentation), a 3M Model 689 printer (C-size drawings) or a Versatec plotter (for roll printing and up to J-size drawings). Once printed, Thaad documentation is registered and checked in and out of the "plot room." This cumbersome, if secure, procedure is mitigated by the ability of users to review documents on their workstations screens, rather than in hard-copy form.
Staff Skills and Applications
For many Thaad employees, TIMS has not required a major skills ramp-up. Rather, it relies on applications already in use like word processing (e.g., Microsoft Word), spreadsheets (e.g., Microsoft Excel), and presentation packages (e.g., PowerPoint). Technical and engineering applications include Pro/Engineer by Parametric Technology (Waltham, MA), I-DEAS by SDRC (Milford, OH); and a variety of Formtek software modules that support process control, application access (at both the network and applications level)
, as well as document indexing, annotation, viewing, redlining, scanning, plotting, format conversion, and printing.
During the two-year development phase of the Thaad program, TIMS is focused on engineering design work. But the system will also be used for the follow-on manufacturing and support phases. "We are using [TIMS] for prereleased data: design specifications, ECRs [engineering change requests], and so on during the design and test phases," said Victorino. "As we progress through the program, we will be developing more product data, more constructions, inspections, and test data. All of this will be coming back into the database."
Managing Engineering Data
Any organization wishing to effectively manage its engineering data--LMSC included--must come to grips with seven base processes that support design and engineering information and its use. These seven key EDMS (engineering data management system) processes are data capture, storage, query, distribution, review and markup, wor
k-flow management, and product focus (see the figure "Engineering Data Management Services"). Most organizations today have these processes in place; the degree to which they are managed and organized depends on the maturity of the enterprise and its management systems.
At a recent Kalthoff EDMS Users Forum (an educational conference for EDMS users and would-be users), Tom Arant, president of Technology Management, a consultancy in Winston-Salem, North Carolina, addressed the issue of EDMS maturity. "Most discrete manufacturing organizations come at an EDMS implementation in three stages or phases," he said. "Phase 1 puts in place the basic engineering data support services: capture, store, query, distribute, and review. Phase 2 expands the scope of Phase 1 and adds work flow to support routing of documents for review and approval. Phase 3 expands basic support services to the enterprise as a whole, extends work flow to all appropriate users, and begins to add product data structures."
In this c
ontext, the TIMS can be seen to be squarely in the middle of Phase 2. Engineering support services are in place and work flow is being used on a limited basis. "Today, all engineering data, analysis, design specs, and textual data change information is collected and routed appropriately," noted Victorino. "Eventually, all of the product data--all the way down through logistics and information required to support [Thaad] out in the field--will be included." Phase 3 data management, in Arant's terms, will incorporate bill of material and purchasing information, as well as other data required to support the overall manufacturing life cycle.
Close Encounters
The Thaad project had a variety of skirmishes with failure, the first even before LMSC was awarded the contract. "We had a requirement for the proposal to be only so many pages in length," said Kadrich. "When we went from System 6 to System 7 on our Mac base, System 7 dealt differently with fonts. As a result, the spreading out of characters in
side the proposal meant that we were about five and a half pages too long." Having spent nearly three weeks cutting the proposal back to its maximum permitted length, Kadrich and his staff were not about to prune further. "We went back to System 6, delivered the proposal, and subsequently won the contract."
Early in the program, a good deal of effort was spent analyzing the processes within the Thaad development program with an eye to both optimizing those processes and developing optimized support systems. "I sat through a number of meetings with my friends in IS [the information systems department] and spent a lot of time trying to analyze the business processes," Kadrich said. On the face of it, this made good sense. But after careful examination, Kadrich and his peers concluded that because LMSC was changing its development approach to reflect its new IPD orientation, as well as implementing a new system that would evolve over time, and because users were not familiar with either IPD or the new sys
tem, the basic business processes would be changing, possibly dramatically. As a result, the idea of "reengineering operations" was shelved until a clearer image of what operations should look like was developed.
A critical component of the Thaad system was an easily mastered interface. "Our user interface is real simple: big buttons, self-explanatory lists," explained Kadrich. Much of the interface was structured to resemble LMSC's E-mail system and users appreciated this commonality. "We were requested in the middle of the program to change our entire E-mail system and our resource scheduling system." Kadrich balked and took his case to LMSC senior management, asking for an exemption to the enterprise-wide mandatory upgrade on the basis of its impact on the project's schedule. His request was granted. "The users did not want to sit down and learn a new E-mail system or how to reschedule resources; they just wanted to do their jobs," he said.
A minor issue that surfaced had to do with voting du
ring design reviews. All members of a review team are required to conclude their examination of documents and vote by the closing date of the review. If one or more members fail to do so, the review cannot be formally closed. But what happens when a team member is out of town or otherwise unable to vote in time? In practice, someone else reviews the materials and votes for the missing person, and the TIMS was modified to support that.
By allowing reviewers to designate an alternate person in those situations where the reviewer is absent, the problem of absenteeism was handled efficiently and effectively. Designated alternates were properly supported by the routing and voting applications.
The Future
While the Thaad program--being geographically distributed and managed more through electronic conferencing than face-to-face encounters--is a bit unusual for LMSC, it's likely to become the norm. Partnering among previous competitors, increasingly decentralized development efforts, and collab
orative programs that push the limits of schedule brevity are going to require systems like TIMS that permit engineering teams to be virtually collocated on a global basis.
This means some big changes. "At LMSC's Fleet Ballistic Missile program, we've pretty much done things the same way for 35 years," commented Victorino. "Because of the short Thaad schedule, we had to think up new ways to do the same good job [as before]." Today, indications are that the TIMS system is doing a better job not only supporting a compressed schedule but also improving product and process quality and cost--improvements that are likely to keep customers coming back.
OVERVIEW
Aggressive schedules are old hat at LMSC (Lockheed Missile and Space Corp.), but the U.S. Army's Thaad missile program was something else. It required an order-of-magnitude improvement in product development cycle time.
Concurrent engineering, or IPD (integrated product development), fit the bill. Rather than a linear development proc
ess, teams with complete responsibility for major subsystems would operate in parallel. If this teaming approach could be successfully coupled with a system supporting on-line conferencing and design reviews, it would be possible to support the program's aggressive schedule.
For its engineering data management system, LMSC needed seamless support of in-use applications and a uniform user interface on Thaad program platforms. Existing classified and secure networks would support in-place desktop systems running a new engineering information management system capable of managing both informal work-in-progress electronic discourse and formal design reviews. Over time, the system would be linked through a secure, encrypted channel to a similar, nonclassified system.
LMSCselected Formtek in late October 1992, after a rigorous qualification process. By February 1993, the vendor had its prototype system in place.
SYSTEMS VIEW
A wide array of platforms and applications are used at LMSCto devel
op design and engineering documentation--everything from a Cray supercomputer to IBM mainframes to Hewlett-Packard and DEC workstations to PCs and Macs. Applications include CAD/CAM, geometric modeling, and finite analysis, as well as database, drawing and presentation, project management, spreadsheet, word processing, and process-planning packages.
FDDI (Fiber Distributed Data Interface) backbones and Ethernet TCP/IP LANs connect nearly 800 users and provide the infrastructure for TIMS (Thaad Information Management System). Based on existing workstations, networks, and applications, TIMS leverages the skill sets of its users and over time, has been contoured to its users' needs.
Primarily a Mac and Unix client/server environment, TIMS supports a variety of engineering data and document management functions. As the Thaad program matures, it will be necessary to link TIMS to legacy manufacturing and purchasing systems.
PROJECT VIEW
A dynamic partnership between LMSCand Formtek supported
the development of the program's keystone application: design review. Without this, the Thaad two-year development program would have taken five years to complete. Because a new development methodology was being put in place, neither users nor managers had a proper foundation for assessing their needs.
LMSCcan now record and archive the product development process, resulting in a chronology of each design's evolution and a repository of practical engineering processes. Because IPDmethodology grants complete responsibility to design teams (including budget and purchase commitments), the ability to play back and analyze the overall development process made LMSC confident that it would be able to contain its costs and meet the objectives of the program.
A variety of problems--ranging from process overmanagement to enterprise-wide changes in E-mail systems--threatened to delay the program, but the Thaad project is on track, and TIMS, its support system, continues to perform above expectation.
Figure: Photograph: James A. Bole, vice president of engineering at Formtek
Figure: IPD Uses Teaming and Technology to Compress Time
Figure: IPD Technology
Figure: Engineering Data Management Services
Scott Wallace is a BYTEtechnical editor. You can reach him on the Internet or BIX at
swallace@bix.com
.
Accelerating Engineering Design
