Bi-Level Integrated System Synthesis (BLISS)
Bi-Level Integrated System Synthesis (BLISS) is an approach that allows design problems to be naturally decomposed into a set of subsystem optimizations and a single system optimization. The BLISS approach overcomes the difficulties of optimizing complex systems such as aircraft or automobiles where multidisciplinary interactions are prominent.
Concept of the BLISS Methodology
BLISS supports design process in which the specialty groups work simultaneously for unprecedented reduction of the total project calendar time. In that process, Group B expects a piece of data from Group A. However, Group B does not wait for that data. Instead, the groups agree by expert negotiation that the data will fall in an estimated range. Now, Group B need not wait for actual data from B. Instead, it generates its responses for several values of the input distributed over that assumed range by a statistical technique of the design of experiments (DOE). All these responses may be generated simultaneously using parallel computing now increasingly available. The responses may reflect local optimization within the group domain.
Next, Group B uses statistical tools to create an approximation (a.k.a. surrogate model) to its responses over the input range and places that approximation in the data base. The other groups do the same and can do that simultaneously because they use the preset input ranges in lieu of waiting for the actual input data. Finally, a system optimization accesses the response approximations data base to find a system design that preserves all the couplings and maximizes the system objective, while all the constraints local to each group have already been satisfied in the group tasks and reflected in the response approximations. The almost instantaneous data retrieval from the data base enables the system optimization to run quickly, which in a sequential process would require convergence of the green/red iteration for every design modification proposed by the system optimizer.
Perform BLISS Using CenterLink
The key requirement of the BLISS solution is to provide user interfaces that help non-experts perform the BLISS MDO (multidisciplinary design optimization) methodology. To accomplish this goal, we combine ModelCenter’s tool integration capability and CenterLink’s Web architecture. In a collaboration scenario of BLISS, there are several groups who are in charge of creating disciplinary analysis models, called black box (BB) models. The task typically involves wrapping applications and connecting them to compute useful metrics of the BB module. Also, a local optimization problem needs to be defined for each BB module. ModelCenter’s integration and built-in optimization capabilities make it an ideal tool to create the BB optimization models. Once each group uploads the ModelCenter BB model to CenterLink, the rest of the BLISS operation is done through CenterLink, running DOE, fitting surrogate models, and performing system optimizations.
BLISS MDO is implemented by combining capabilities of CenterLink and ModelCenter. In the collaborative scenario, groups use ModelCenter to create BB optimization modules. Each group can select an optimization tool that works the best for the BB module. The subsystem design groups upload the models to CenterLink. When the BB module is uploaded, the BB group proceeds to perform DOE and to generate surrogate models using CenterLink. When surrogate models of all the BB modules are available, they are used by the system optimization.
The BLISS process was implemented through Web interfaces of CenterLink. CenterLink serves as a Web portal for engineers who are performing design optimization. Required tasks for BLISS are assigned to users based on their roles. Users perform the entire BLISS process by following eight steps defined.
- Step 1: Define BLISS Project
- Step 2: Select Users
- Step 3: Define Variables
- Step 4: Define System Objective
- Step 5: Define Variables Bounds
- Step 6: Execute BLISS Process
- Step 7: View Design History
- Step 8: Control BLISS Process

