The Engineering Mindset Part 6: Complex or Complicated? Measurements and Targets – be careful what you ask for

Chris and Penny Hamlin explain how real-time data and dynamic insights can drive sustainable change

CURRENT approaches to driving change focus on carrots and sticks, incentives and penalties, using measurements that are rooted in linear concepts. Encouraging positive actions in complex systems, whether designing petrochemical plants, running existing facilities, establishing corporate goals or purchasing energy, requires a more sophisticated approach. Intriguingly we can learn a lot from exploration of the “Red Planet”.

Think about the contrasting ways in which NASA engineers drive their cars to work and how they drive the Perseverance Mars rover. Perseverance has a top speed of only 152 metres per hour, but operating the rover is one of the most complicated and time-consuming tasks of the mission. With six wheels, four steering motors, a 2.1 m robotic arm, and 23 cameras and two microphones acting as eyes and ears, the level of diligence and precision applied by NASA engineers to manoeuvre a vehicle that is, on average, 225m km away is remarkable.

In contrast to driving to work “on autopilot” engineers operating Perseverance have to first process historic data, use this to inform and validate new objectives and create detailed execution steps. These are then transmitted to Mars with no direct oversight or ability to adjust. They then wait, often overnight, to find out what happened before planning the next steps.

These contrasting tactics illustrate the latter two of these three common target-setting approaches:

1. Standards-based – universal, historic and assumed. Industry standards tend to be aggregates or averages that typify the measure. While they have their uses, the inability to differentiate is problematic and they can only change if large parts, or all, of the industry adopts a new approach.

2. Retrospective, actual – periodic, historic and often aggregated, derived or inferred. These lagging measures are too late for proactive management and tend towards costs and penalties. Feedback is delayed. This is how Perseverance is operated.

3. Run-time visibility, actual – continuously calculated, measured, or derived. The targets are local and bottom-up measures. This means that information is immediate, verifiable, and dynamic which enables proactive steps to be taken when issues arise. This reflects how we all drive to work and back every day.

When considering complex issues such as mitigating greenhouse gas emissions, the more visible and current the measurements are, the greater the impact you can have.

Current state

We manage facilities like NASA operates Perseverance, relying on a structured process: collecting data, validating it, defining objectives, and deploying plans with no direct oversight. For Perseverance, this method is necessary due to interplanetary communication delays and bandwidth limitations – it can take between three and 22 minutes for a signal to reach the rover. However, on Earth, this approach is increasingly outdated.

Despite advances in digital capabilities, businesses and governments still rely on standards-based metrics, retrospective reporting, and aggregated KPIs. These traditional methods fail in complex systems, where simple definitions and measurements are inadequate.

The problem with retrospective data

The C-THRU project examined global emissions data related to petrochemical production, revealing inconsistencies in life-cycle assessments (LCA) that obscures true environmental impacts. The attempt to impose order on a complex problem by averaging data and assuming homogeneity leads to misleading conclusions and ineffective policies. Figure 1  shows how variable the emissions intensity factors are for individual chemicals across different datasets and processes. This demonstrates the inadequacies of standardised, retrospective metrics in driving change.

The C-THRU team found that global emissions estimates for petrochemical production carry an uncertainty of around 34%, with significant variations stemming from missing facility-level process data, inaccurate reporting of feedstock emissions, and unverified assumptions in life-cycle assessments. This highlighted the need for improved facility-specific emissions data and real-time monitoring to reduce uncertainty and enhance decision-making accuracy. The team also demonstrated that by prioritising facility-level process specification in data collection for just 20% of facilities, global uncertainty could be reduced by 80%.