Explore the Design Space before Designing!
Most MDAO processes start with rough sketches, then flesh that out into one more models parameterized to support the desired design space exploration, with enough detail to feed the various analysis tools that they want to wire up into the MDAO workflow, which then generates the visuals of the design space that the engineers get to explore.
The problem with that, however, is that design space exploration is coming after they have made a great number of decisions in creating the rough sketches and the parameterized models. Further, those many decisions are effectively “baked into” the design space visualizations. We really want to give the engineers visibility to the entire design space; and we want to give them that visibility prior to making any decisions (so, prior to creating any rough sketches, let alone parameterized models).
That’s where set-based analysis can be such a valuable addition to this process! Rather than starting with point-based sketches driving an MDAO workflow through a series of point-based analysis tools, we can inject a set-based analysis step ahead of that, generating the same sort of design space visualization for the engineers to use to narrow the sets under consideration, optimizing much of the design before they generate the first sketch.
The net result is that much more of the design space is explored, leading to far more innovation and far more optimal initial sketches feeding your MDAO effort which is then doing its point-based optimization in more preferable portions of the overall design space.
But that is just one of the many benefits…
Superior Visual Collaboration on the Models!
Those experienced with MDAO-based processes will often point to disconnects between the MDAO engineers and the engineers in the various disciplines, including the systems engineering discipline.
Consider a typical aircraft project, where the systems engineers are managing various requirements on the weight, payload capacity, speed, endurance, and cost of the aircraft; and they may have the system design captured as a collection of SysML diagrams. Each of the discipline engineers (e.g., aerodynamics, propulsion, structures, mass properties, guidance and control, electrical, etc.) will have their own discipline-specific diagrams that will each have specific sub-weights, sub-costs, sub-capacities, and speeds. When pulling all those together into models, the MDAO engineers know they must be very careful to properly map the values that each of the discipline engineers are talking about. For example, the weights don’t just all add up into one total weight; the weights have different meanings and different timings. For example, there is the Takeoff Gross Weight and there is Weight Empty — knowing what goes into each can have a critical impact on the modeling. Similarly, there are many different speeds… for example, there is the air speed, the speed relative to the ground, the rate of vertical climb, and so on. Further, you need to make sure the assumptions being made in each model are consistent before you map quantities as if they are the “same”.
If you have questions, would like a demo, or would like to schedule us to visit you, please email us at Answers@TargetedConvergence.Com. Alternatively, you may call us at 1-888-LRN-FRST (1-888-576-3778). Either way, we'll route you to the right person.
Or if you prefer, you can send email direct to:
Our Sales Team at Sales@TargetedConvergence.Com.
Our On-the-Job Coaching Team at Mentors@TargetedConvergence.Com.
Our Software Support Team at Support@TargetedConvergence.Com.
Our Website Team at Webmaster@TargetedConvergence.Com.
Our Accounting Team at Accounting@TargetedConvergence.Com.
Our Human Resources Team at HR@TargetedConvergence.Com.
After 10 years in Carrollton, we have moved a few miles south to the Dallas Communications Complex in the prestigious Las Colinas region of Irving, TX:
Targeted Convergence Corporation
400 E Royal Ln Ste 290 Bldg 3
Irving, TX 75039-3602
Success Assured® starts with a Causal Map… a diagram intentionally kept very simple (just 4 shapes to learn) so that you can pull in engineers from any discipline and have them immediately offering insight into what’s missing, what’s the same, and what assumptions need to be considered. That Causal Map then evolves into an even simpler Decision Map that can be used to build set-based Trade-Off Charts that provide visibility to the overall design space. That combination of Decision Map and Trade-Off Charts gives the engineers visibility to both the structure of the design space and its limits and sensitivities. That allows the disconnects between the disciplines to get identified and ironed out before the first sketches are generated, let alone before the first MDAO workflows are constructed.
In fact, a surprising level of analysis, innovation, and optimization can be done before the first MDAO computations have been done. Often even before the first set-based computations have been done… the Causal Map has been known to lead to such clarity that significant portions of the design space can be “eliminated as weak” based simply on the knowledge made visible by the Causal Map. Yes, that would have eventually been exposed by the MDAO effort… but not before tremendous work may have been performed that wasn’t needed.
Added Benefit: Success Assured® is Omni-Directional
Point-based computations are performed in a particular direction. For example, consider Excel where each cell computes a value from a formula of the form “= f(x, y, z)”. If you later decide you want to compute y from that cell and x and z, then you need to rework the algebra in another cell. (Or you need to setup an algorithmic solver to run that spreadsheet repeatedly with different values.)
Similarly, the MDAO workflows are computed in a specific ordering based on what tools compute what outputs from what inputs. Those tools operate in a particular direction. (Similarly to Excel, you can use the MDAO optimizers to run those repeatedly to effectively compute backwards.)
In contrast, when you map out the relations in Success Assured®, those relations are omni-directional. You can write the equation once as “P V = n R T” or “x^2 + y^2 = log(z)” or similar, and then you can decide later what you want to compute from what, without needing to rework the equation. That makes the models far more reusable and far more flexible for design space exploration. That allows you to decide what you want to compute after you’ve mapped the model out and after you’ve already done some design space exploration! No need to re-work the model to compute that different thing; just specify what you want to see, and Success Assured® handles the computation re-ordering.
That can have huge benefits in accelerating the development of your MDAO models because it allows you to capture the knowledge generically just once, and explore the design space before choosing which tools to execute in what direction, which will then drive the implementation of your MDAO workflows, and the parameterization of your models.
Contact Us to Schedule a Demo today!
Do you have some first order models you use on the front end of your MDAO efforts? If you can share those with us, we can even demo using your own data!
Confidently make decisions you won't need to change...
Does your team use MDO or MDAO* in your conceptual design work?
If so, adding Success Assured® can make your MDAO better!
(We explain a few of the ways below.)
* If you are wondering what MDO and MDAO stand for, then this is not the web page for you; instead, please visit our page on Better Multi-Dimensional Trade Studies.
Added Benefit: Sets Don’t Miss Anything
When running a DOE, a trade is often required between how many runs are being made and how good the coverage is of the design space. More runs means it is less likely you miss anything; however, more runs requires more time OR requires lower fidelity models, which can result in you missing things due to the lower fidelity.
The engineers typically want full factorial. Worse than that, for non-linear relationships, they want to increase the number of levels on which they are computing that full factorial.
Set-Based Analysis is effectively doing full factorial on ALL the levels. For continuous quantities, that is sets of an infinite number of points formed by all the combinations of the infinite values of those continuous quantities. How is it possible to compute an infinite number of points in finite time? By computing in terms of sets, not individual points. That has the advantage that there is no concern that if you picked a different increment or different set of factorial values that you’d discover a portion of the design space that your current analysis missed.
With set-based analysis, it only eliminates the portions of the design space it can prove are infeasible; but if a portion has been eliminated, then you don’t need to ask how many more runs you need to do to be confident there is not a better answer that you just haven’t discovered yet.
Similarly, unlike point-based methods, you don’t have to worry about functions with sharp peaks or valleys being missed, or functions with varying oscillations being misinterpreted. The set-based analyses won’t miss any of those. (For example, the charts to the right are showing a fairly linear relationship that has a sharp peak at a particular X value. But if you plot it in Excel, you may never see that peak... unless you generate your chart from thousands of repeated rows such that you get lucky and hit the sharp peak. No need to worry about that when charting Set-Based.)
Of course, those advantages come with disadvantages. But that’s what makes set-based Success Assured® such an effective complement to your point-based MDAO: together you get all the advantages of both, filling in for the disadvantages of each.
Reversing the Flow and Doubling the Return!
In case it wasn’t clear from the prior discussion, we are not proposing that Success Assured® is at all a replacement for MDAO. Rather, we are suggesting you inject Success Assured® in between steps 1 and 2 of the MDAO process: use Success Assured® to further flesh out the Problem Definition, constructing a Causal Map representing that Problem Definition, then evolve that into a Decision Map that can guide the rest of the MDAO process, evolve that Decision Map into set-based Trade-Off Charts that give visibility to the larger design space, before you start making the decisions that go into the 2D sketches and 3D models that eventually feed the MDAO effort.
But let us take that one step further, showing how you can further leverage that MDAO investment…
In evolving the Causal Map into a computable Decision Map, you will need to construct set-based Relations between the various decisions that need to be made. Very often those are constructed as response surfaces based on running DOE’s on more detailed models or over test data. Sound familiar? Yes, that MDAO infrastructure can be employed to feed the set-based models that we are suggesting you inject in the front end of your larger MDAO process. Of course, those will be very different MDAO models… simpler workflows, modeling smaller subsets of the overall system. And once captured, that will often be reusable since it is generic independent subsystem knowledge.
In that sense, you are doubling the return of your MDAO investment, by introducing a reverse flow where you use the MDAO DOE capabilities to generate response surfaces from your point-based tools, that can be then be leveraged in set-based analyses to explore the overall design space prior to doing the first sketches (where you make the first point-based decisions for this new project), which then go into your normal MDAO workflows.
Often when we deploy Success Assured® in organizations that do not do MDAO, we need to teach them to use the MDAO frameworks (or similar) to generate the response surfaces that we need to do the set-based analyses. Such use of those MDAO frameworks are relatively simple, but still far superior to the alternative.
Added Benefit: Success Assured® covers broader Project Planning
The Success Assured® Causal Map has explicit support for doing the type of project planning you might use MS Project or PERT to do, extended to cover project costing decisions, but further showing the impacts that product design decisions may have on those costs and timings. In that way, your trade space may include the project timing and costing decisions that are being traded off against the performance of the product that the project is designing.
Often those trade-off decisions need to be made very early in the project (during project planning), long before you can assign the resources to perform the MDAO, let alone actually perform the MDAO to make visible that trade space. Success Assured® can allow you to create those trade space visualizations to properly inform that decision-making.
Further, often there’s a great deal of uncertainty during project planning, which the set-based analyses can trivially model. Similarly, risk management is often a significant portion of the project planning effort. All such uncertainty and risk elements can be handled by the Success Assured model, keeping them out of the complementary MDAO models.
Then later in the project, as the MDAO results start to become available, those can be used to adjust the Success Assured® models and Trade-Off Charts to then give visibility to how the project plans need to be adjusted based on the new knowledge that has been made visible.