Challenges in Making the Transition to Digital Engineering


“Digital acquisition holds the important thing to unleashing the pace and agility we have to discipline functionality on the tempo required to win in a future battle with a peer competitor,” wrote Darlene J. Costello, appearing assistant secretary of the U.S. Air Pressure (Acquisition, Expertise, & Logistics) in an Air Pressure memo in Might 2021.

Making the transition from conventional improvement practices to digital engineering (DE) presents challenges. This weblog publish describes a few of the challenges I noticed whereas working with the U.S. Military Capstone Future Vertical Elevate (FVL) program. In Capstone, I used to be particularly searching for prices and advantages of model-based techniques engineering (MBSE) within the context of introducing the architecture-centric digital integration course of (ACVIP) utilizing the Structure Evaluation and Design Language (AADL). These observations embody transition challenges inside the rising digital engineering ecosystem.

The Promise of Digital Engineering

Tom McDermott of the Programs Engineering Analysis Heart defines digital engineering as

an built-in digital method that makes use of authoritative sources of techniques knowledge and fashions as a continuum throughout disciplines to help lifecycle actions from idea via disposal.

I’ll return to the important thing ideas from this definition of built-in and continuum throughout the lifecycle on this publish. After all, the eponymous attribute is that digital engineering makes use of digital instruments and representations within the technique of growing, sustaining, and sustaining techniques, together with necessities, design, evaluation, implementation, and check. As described within the SEI weblog publish “What Is Digital Engineering and How Is It Associated to DevSecOps?,” digital engineering is effectively suited to the DoD’s have to maintain and keep long-living techniques whose missions evolve over time. The digital modeling method is meant to determine an authoritative supply of fact (ASOT) for the system during which discipline-specific views of the system are created utilizing the identical mannequin components. This model-based method carries ahead into the design and implementation.

A digital modeling surroundings successfully applies MBSE to the design and creates a typical standards-based method to documenting a system that enforces the usage of requirements by all stakeholders. A typical modeling surroundings with generally accepted and well-defined properties and stereotypes is meant to enhance the power to research the system and cut back the probability of discovering late defects. The provision of digitized system knowledge for evaluation throughout disciplines ought to present constant propagation of corrections and incorporation of latest data.

Ideally, with MBSE this data may be acknowledged as soon as after which robotically propagated to numerous views of the info for all stakeholders. The results of this method is an general discount of improvement dangers, the power to search out and proper defects earlier in improvement when adjustments are comparatively cheap, and elimination of document-driven improvement. I attempted to measure this profit in Capstone.

The foreword to the DoD Digital Engineering Technique of June 2018 summarizes the anticipated enhancements to system improvement and upkeep that the DoD envisions for its techniques:

…incorporating the usage of digital computing, analytical capabilities, and new applied sciences to conduct engineering in additional built-in digital environments [will] enhance buyer and vendor engagement, enhance menace response timelines, foster infusion of expertise, cut back value of documentation, and impression sustainment affordability. These complete engineering environments will enable DoD and its trade companions to evolve designs on the conceptual section, decreasing the necessity for costly mock-ups, untimely design lock, and bodily testing.

Nonetheless, measuring the effectiveness of digital engineering/MBSE has been elusive. Though the DOD digital engineering technique emphasizes the conceptual section, designs should be captured because the ideas evolve. We centered on this undertaking on digital engineering for embedded techniques via ACVIP. The ACVIP course of is constructed on a digital engineering methodology and AADL-based instruments implementing the ideas expressed within the Programs Modeling Language (SysML). Engineers due to this fact used a various toolset throughout the event phases of the software-development lifecycle for embedded computing techniques. Whereas AADL is exact in semantics and items, the SysML used within the conceptual section is looser. Some engineers thought that the dearth of precision on the boundary between abstraction layers made it arduous to share and analyze the fashions.


We encountered a number of challenges associated to confounding from a number of instruments and requirements, the educational curve, and measurement.

A primary problem was that ACVIP was not the one new method launched. This system launched different digital engineering instruments and requirements through the undertaking together with, amongst others, SysML, the Future Airborne Capabilities Setting (FACE) knowledge mannequin, Complete Structure Technique (CAS), Enterprise Course of Mannequin and Notation (BPMN), Unified Profile for DoDAF/MODAF (UPDM), and related instruments.

A second problem was that the engineers had been nonetheless studying learn how to apply the strategies for each ACVIP and MBSE. A 3rd problem was that measurements didn’t isolate course of components sufficiently to measure the causal results of particular strategies. Consequently, though we confirmed some optimistic advantages, we couldn’t attribute them to particular instruments or strategies.

To raised perceive the challenges, we not solely reviewed efficiency knowledge but in addition surveyed individuals to get their subjective opinions on digital engineering and MBSE effectiveness. The survey requested quite a lot of questions, together with demographics; earlier expertise within the area; expertise with the MBSE applied sciences used on Capstone; help for the event surroundings; and the perceived effectiveness with respect to value and length. A complete of 26 folks responded, two from authorities acquisition and the remaining about evenly cut up between improvement and integration roles.

Recurring themes within the survey outcomes included

  • an absence of expertise with digital engineering instruments and practices, and with ACVIP and MBSE
  • restricted instrument maturity
  • substantial value of entry to be taught and use a few of the instruments
  • restricted course of maturity, with no built-in end-to-end course of, as evidenced by lack of overarching measurement and metrics
  • want for dedication to an overarching technique with follow-through
  • level options that didn’t work successfully collectively
  • points with coaching, tooling, and expertise

Few people surveyed had earlier expertise with MBSE, ACVIP, FACE, and digital engineering, and so they steadily cited rising instrument limitations as a typical obstruction. Even the place the group had used digital engineering, few people labored on the sooner undertaking and few established practices had been carried over to the brand new undertaking. Open-ended feedback within the survey referred not solely to inadequate coaching, but in addition to the failure of contractors to combine the brand new practices into the event course of. To be efficient, due to this fact, the brand new instruments and practices wanted further transition help.

Integrators—those that assembled {hardware} and software program parts—had been extra optimistic concerning the optimistic schedule results of utilizing MBSE than part builders. The part builders surveyed agreed that ACVIP shortened the length to integration or check, whereas no integrator thought that ACVIP elevated the length. An inexpensive rationalization for this divergence of opinion is that builders see solely the early prices and further efforts, whereas integrators see the advantages of the additional efforts to reach at an improved state prepared for integration. This divergence additionally illustrates why it’s crucial to have a holistic view of the event reasonably than to depend on people who may micro-optimize on the expense of general system efficiency.

Our survey had too few responses for us to claim scientific validity, however we do assume that the outcomes recommend some drawback areas that warrant additional investigation. Though subjective opinions are usually not essentially true and don’t guarantee a broad perspective, they do seize a few of the frustration felt by a minimum of a few of the working engineers.

Points with Instruments

Lots of the survey respondents pointed to lack of any integration throughout MBSE practices, and that drawback was confounded by the introduction of a number of digital engineering instruments and strategies. Instrument points included the method being too handbook and lack of workflow coordination among the many instruments supplied. This survey can’t decide particular instrument or course of issues or learn how to resolve them. Nonetheless, a number of respondents felt underprepared or inadequately supported.

We noticed that the toolchain linking necessities utilizing SysML to data-modeling instruments utilizing the FACE normal and ACVIP had limitations. Fashions and gear help didn’t stream seamlessly between the completely different abstractions. This inconsistency weakened the power to have an ASOT and inspired working at an inappropriate stage of abstraction.

These instruments had been underneath improvement and improved significantly throughout Capstone execution. There are two separate classes from this commentary. First, integration of software-systems approaches right into a coherent workflow is a key enchancment that might facilitate adoption of digital engineering. Second, whereas measurement ought to help course of evolution, a secure course of and tooling are wanted to measure value effectiveness.

The desire for a instrument that crosses ranges of element may recommend that there is no such thing as a clear plan throughout the lifecycle for growing and utilizing fashions. When requested, “How will this mannequin be used going ahead?” the builders didn’t present a solution. We nonetheless see contractors current fashions in PowerPoint, which signifies that though SysML clearly fashions structure, construction, and conduct precisely as meant, some desire to mannequin in Visio or PowerPoint to elucidate what they did. This reliance on PowerPoint turned much less frequent as Capstone progressed, and generally the Powerpoint transcribed SysML or Division of Protection Structure Framework (DoDAF). Whereas sluggish instrument adoption could also be a problem with the instruments themselves, coaching, expertise, or mentoring, it exhibits that simply offering instruments or mandating their use shouldn’t be sufficient. These answerable for the transition to digital engineering ought to anticipate that there can be a interval of some length throughout which the customers grow to be accustomed to a brand new approach of working.

A consultant free-form response was that “It’s arduous to level to any tangible profit…mannequin creation is basically handbook; our program has struggled to maintain up with modeling. Something discovered has all the time been 2-to-3 steps behind the purpose the place it will have been helpful as a result of we merely don’t have the sources to manually duplicate modeling.” This response is a sign of inadequate coaching mixed with tight execution timelines.

This response additionally neatly encapsulates problems with expertise, instrument integration, and the dearth of a coherent organizational course of—all key impediments to the profitable adoption of digital engineering. A technology-transition program ought to allocate time to be taught whereas requiring incorporation of the brand new expertise into the modeling method.

Constructive Recommendations for Measurement and Enchancment

Packages want the improve functionality that digital engineering supplies. However digital engineering represents a large-scale change requiring a complete funding plan for expertise transition. Limitations to efficient digital engineering embody instrument functionality, personnel expertise, and general administration. If you’re answerable for the transition of this expertise, you have to begin the advance course of with clear statements of the objectives and observe up with a sensible measurement program to judge progress towards attaining these objectives.

The important thing goal for our examine involved measurement. How are you going to inform if what’s being executed is an enchancment? A number of course of introductions and adjustments throughout execution confound this analysis. Lack of adequate experience utilizing the brand new course of makes it arduous to know learn how to measure success successfully, as a result of many steps in adoption require a distinction between whether or not the method is ineffective or whether or not it’s simply not getting used successfully. Specifically, first efforts mirror the early phases of a studying curve, and thus might not symbolize a good check.

Measurement would even be made simpler by narrowing the scope of digital engineering introduction in order that you can consider every step in adoption extra fastidiously. After all, narrowing scope would create different dangers. Nonetheless, quick, large-scale change is seldom profitable. Together with change-support brokers or mentors who already know the processes can also be important.

One other profit to narrowing the introduction scope and including experience via mentoring or just-in-time help is that it makes potential a concentrate on satisfying wants. As workflows are developed, you need to maintain the digital engineering artifacts at applicable ranges of abstraction. Coaching adopted by lively mentoring helps to attain the appropriate stage of abstraction, whereas the problems with the handoff from one position to the subsequent make clear the abstraction wanted. To realize an ASOT, you have to guarantee that the digital engineering artifacts evolve and are literally used. Each circumstances—evolution and precise use—are observable and will point out success or alternatives for enchancment.

It’s also crucial to introduce new workflows the place the digital engineering implies one thing completely different from the outdated approach of working. A lot of the facility of ACVIP is in enabling digital integration and digital evaluation earlier than shifting to bodily parts. You could combine these actions right into a coherent incremental workflow, and make an effort utilized and the rework found seen. It’s good if ACVIP finds points in digital evaluation reasonably than in bodily check, and higher if ACVIP prevents these points from rising solely. Solely by monitoring the trouble and rework in digital and bodily integration and check will you realize what impact ACVIP has had.

You must design and refine metrics to measure the expertise and course of. With respect to ACVIP in advanced embedded techniques, rework is the foremost value subject addressed. You must outline, isolate, and monitor rework prices within the context of when points had been found, and outline subject discovery itself in order that it may be captured constantly. Solely after measurement is efficient can it’s used to tune the method, thus maximizing return on funding and decreasing integration dangers. In the identical approach, extra generalized MBSE would want focused metrics.

Lastly, digital engineering ought to present an ASOT, i.e., the digital engineering merchandise must be utilized by most stakeholders throughout the event to speak and evaluate. Demonstrating ASOT was not the aim of Capstone, however a few of the points advised that the interfaces and handoffs had been lower than splendid. The gathering of particular person instruments should perform inside an general ecosystem. It is very important guarantee that instruments have full data, however on the applicable ranges of abstraction. It’s also fascinating that the outcomes from one instrument may be transferred to a different, and ideally again once more.