The Hidden Cost of Over-Engineered Simulations
By Alejandro Elizondo
The notion that "more" equates to "better," that increased effort yields superior results, and that more complex solutions are inherently more valuable is a common belief. While this may hold true in some contexts, it does not apply universally, particularly in the realm of process simulation.
When planning to initiate, quote, design, or propose a simulated solution, it is essential to begin by posing the following critical questions. When project costs begin to spiral out of control, it is often a direct result of failing to answer these questions properly, leading to hidden, unanticipated expenses.
What are the specific requirements?
What is the intended purpose of the simulated environment?
What is the desired timeline for implementation?
What information is necessary to build this simulation?
What budget constraints exist for this project?
There is a prevailing tendency to assume that a high-fidelity solution can address all necessary aspects of a project. For this reason, it is important to understand the hidden costs typically related to over-engineering a project:
Data Scarcity: The effort required to force models to converge when critical data is incomplete or unavailable.
Unnecessary Complexity: The time and resources consumed by creating overly complicated models to reach a level of high fidelity that is not truly needed.
Schedule Delays: Tight schedules that prevent the time needed to properly test and validate high-fidelity models.
Effort Underestimation: The misjudgement of the effort required for a high-fidelity model, leading to significant time overruns.
Increased Maintenance: The fact that changes in the plant require updates in the model, and the higher the fidelity, the more time is needed for these updates.
To avoid these hidden costs, several considerations must be evaluated when contemplating the development of a high-fidelity model:
Is the requisite data available to create a high-fidelity model?
Are sufficient time and financial resources allocated for the development of such a model?
Is there a necessity for a model that operates in real-time to execute all plant processes?
Do the project objectives genuinely necessitate a high-fidelity model?
Typically, these questions are overlooked, and what is often required is a simulated environment capable of providing accurate responses to key variables (such as pressure, temperature, flow, and level) within an acceptable timeframe. It is crucial to maintain a focus on actual needs and to prioritize the core requirements of the project. A simplified model can be a great starting point, allowing for expansion as a deeper understanding is attained.
In conclusion, the principle that "more" is always better does not hold true in every scenario. It is vital to stay grounded in the current requirements, maintain perspective on project necessities, engage in the development process, and strive to maximize the utility of the simulated system. By doing so, stakeholders will be better positioned to identify further needs and refine their understanding of the project landscape.