AI and Human-Centered Skills in Modern Engineering

Natural-language models powered by generative artificial intelligence (AI) technologies can enhance human-centered skills, such as communication, emotional intelligence and interpersonal workflows in chemical-manufacturing settings

The chemical process industries (CPI) operate in environments defined by nonlinear system behavior, tight safety margins and significant economic and environmental consequences for accidents. Engineers must not only perform rigorous technical analysis, but also communicate risk clearly, collaborate across disciplines and make decisions with uncertainty, as emphasized in Ref. 1 and reinforced by the National Academy of Engineering [2] and the AIChE Competency Framework [3]. Artificial intelligence (AI) is becoming a practical enabler of these human-centered skills. In modern refining, petrochemical and energy-transition facilities, AI tools support clearer communication, more structured reasoning, improved time management and stronger ethical judgment.

Rising system complexity further increases the need for these capabilities. Engineers must interpret large volumes of real-time data, respond to rapidly changing operating conditions and coordinate decisions across organizational and technical boundaries. Traditional workflows remain essential, but they are no longer sufficient on their own. AI systems help bridge this gap by providing faster access to information, clearer explanations of process behavior and more consistent decision support during abnormal situations.

Workforce dynamics add another layer of urgency. Many organizations face simultaneous retirements of highly experienced engineers and the onboarding of younger professionals who bring strong analytical skills, but limited exposure to complex plant scenarios. AI can help reduce this experience gap by capturing institutional knowledge, generating clearer documentation and supporting more structured communication between senior and junior engineers. These capabilities strengthen both technical performance and the interpersonal competencies that underpin safe and reliable operations.

AI also enhances situational awareness by summarizing operating logs, highlighting deviations and identifying subtle correlations in process data. This shift from reactive to proactive decision-making supports a more resilient operating culture and reduces the likelihood of human-factor-related incidents. As the digital transformation accelerates across refineries, petrochemical plants and energy-transition facilities, engineers who integrate AI into both technical and interpersonal workflows will be best positioned to lead [4]. Understanding how these tools enhance human centered skills is therefore essential for the next generation of engineering excellence.

Categories of AI tools

AI tools used in the CPI can be understood through several functional categories that reflect how engineers interact with digital systems in daily practice. Table 1 summarizes the major categories, representative tools and typical engineering applications, reflecting the digital transformation trends described by McAfee and Brynjolfsson [5] and recent industrial analyses [4]. Figure 1 depicts how AI tools deliver piping and instrumentation diagrams (P&ID) tools for process flow diagrams (PFDs), emphasizing that early-stage engineering documentation supports process understanding.

Predictive analytics systems process extensive process data to discern intricate multivariate relationships and spot warning signs of equipment decline or process disturbances well in advance of them leading to operational difficulties [4]. Optimization and simulation AI tools support process design, debottlenecking and energy-efficiency studies by combining first-principles models with machine-learning algorithms to accelerate scenario evaluation. Natural language models improve communication clarity, rewrite technical documents, summarize operating logs and translate complex engineering concepts for cross functional teams. Collaboration and workflow AI tools streamline project coordination by generating meeting summaries, extracting action items and organizing communication threads across disciplines. These tools enhance communication by assisting engineers in improving written messages, detecting unintentional severity or vagueness and bolstering leadership impact in crucial safety settings [6].

Strengthening soft skills

AI plays a significant role in improving communication among CPI professionals by helping engineers produce clearer operating procedures, safety bulletins, management-of-change (MoC) documentation and HAZOP responses. Natural-language models identify ambiguous terminology, propose more precise alternatives and generate visual aids, such as risk matrices and flow diagrams, which support design reviews and safety briefings. Visualizing likelihood-versus-consequence relationships becomes far more consistent when engineers use digital tools that generate structured risk matrices, supporting clearer decision making during abnormal or uncertain operating conditions (Figure 2).

Engineers can prompt these tools with requests such as, “Rewrite this safety memo for operators and highlight the consequences of exceeding the temperature envelope,” or “Explain this interlock sequence in simpler language suitable for non-technical stakeholders.” These interactions help engineers refine their communication in ways that reduce misinterpretation and improve operational clarity, consistent with the communication principles described by Felder and Brent [7].

Decision making and critical thinking also benefit from AI, as predictive analytics highlight correlations that may not be visible through manual review, present alternative interpretations of data and document the rationale behind engineering decisions. Refining how engineering messages are interpreted becomes easier with tools that analyze tone and sentiment, helping teams communicate more clearly, respectfully and effectively across technical and operational groups (Figure 3).

Engineers can ask, “Identify any correlations between regenerator temperature fluctuations and air blower performance,” or “Simulate the impact of reducing catalyst circulation by five percent.” These capabilities help engineers recognize cognitive biases, evaluate trade-offs more systematically and make more defensible judgments under uncertainty, reflecting the decision science insights of Kahneman [8] and the systems thinking approach of Leveson [9].

Emotional intelligence is strengthened through AI driven tone-analysis and conversational-simulation tools that allow engineers to rehearse difficult discussions, refine phrasing and anticipate how their communication will be perceived by colleagues, operators or regulators. This aligns with Goleman’s foundational work on emotional intelligence [6] and Edmondson’s research on psychological safety [10]. Time management improves as AI reduces cognitive load by summarizing daily operating logs, retrieving historical incident reports, organizing project files and prioritizing tasks during turnarounds or major design projects. Ethical reflection is also enhanced, as AI prompts engineers to consider environmental impacts, regulatory compliance, safety-versus-throughput trade-offs, and long-term sustainability implications, consistent with Shneiderman’s principles of human centered AI [11].

Digital tools in practice

AI’s practical benefits are best understood by observing how engineers integrate these tools into their daily work. In safety communication, for example, an engineer drafting a memo for a new distillation train may begin by writing a preliminary explanation of operating limits and protective systems. Clearer safety communication during distillation train design is achieved when engineers use digital tools that translate complex operating limits and protective system logic into more accessible explanations for operators and reviewers (Figure 4).

A natural language model then reviews the text, identifies ambiguous terms and suggests clearer phrasing. Engineers often use prompts such as, “Improve the clarity of this explanation of high-temperature trip logic,” or “Generate a simple risk matrix for this operating envelope.” Predictive analytics generate a risk matrix that visually communicates the relationship between likelihood and consequence, consistent with Center for Chemical Process Safety (CCPS; www.aiche.org) guidance on risk communication [12]. The engineer may then use an AI-based roleplay tool to simulate regulatory or community questions, refining explanations to ensure clarity and empathy. A tone-analysis tool provides final feedback on the emotional impact of the message, resulting in a more effective and accessible communication package.

In troubleshooting of fluid-catalytic-cracking units (FCCU), AI ingests regenerator temperature data, air-blower performance metrics and catalyst circulation trends. Multivariate models identify correlations between coke formation and temperature fluctuations that may not be immediately apparent. More consistent and proactive operational decisions become possible when engineers use digital decision support tools that synthesize real-time refinery data, highlight deviations and prioritize corrective actions (Figure 5).

Engineers can prompt the system with, “Analyze this FCCU dataset and identify any leading indicators of temperature instability,” or “Simulate the effect of increasing air-blower capacity by three percent.” Scenario simulations allow the engineer to test corrective actions, such as adjusting air blower settings or modifying catalyst circulation rates. The system then generates a decision log that documents the rationale behind each option, the associated risks, and the expected outcomes. This structured approach strengthens both technical analysis and communication with operations and management, reflecting best practices in process safety and reliability engineering [13].

In leadership and cross-functional collaboration, AI tools summarize meeting transcripts, extract action items and highlight areas where communication may have been unclear or emotionally charged. Engineers can request, “Summarize this meeting into action items for operations, mechanical, and safety teams,” or “Identify any emotionally charged language in this email and suggest improvements.” This improves delegation, reduces meeting duration and enhances team cohesion by ensuring that all participants share a consistent understanding of priorities and responsibilities.

To illustrate how AI can support this workflow, a simple example involving the computer programming language Python is shown here. The script uses an anomaly-detection model to identify abnormal regenerator-temperature behavior correlated with air-blower pressure drops and catalyst-circulation instability. This type of analysis can be performed using standard engineering datasets and can be run in Visual Studio or any Python environment.

A complete, self-contained Python program demonstrating this analysis is provided in Appendix B for readers who wish to reproduce the results or adapt the workflow to their own FCCU datasets.

Case studies

A refinery operating at 200,000 barrels per day (bbl/d) implemented AI-enhanced communication tools during the design and commissioning of a new atmospheric distillation train. The process engineers responsible for safety documentation used natural-language models to simplify technical language and predictive analytics to generate risk matrices that clarified operating envelopes. Prompts such as, “Rewrite this explanation of the reflux-drum high-level alarm for operator clarity,” helped refine communication. As a result, clarification requests from operators decreased significantly, regulatory review cycles accelerated and safety audits revealed fewer misinterpretations of critical information. These improvements align with established principles of effective safety communication [12, 13].

In a Middle East refinery operating a 45,000-bbl/d FCCU, intermittent regenerator temperature fluctuations prompted a detailed investigation supported by AI. Predictive analytics identified correlations between air-blower performance, coke formation and temperature spikes that were not evident through manual review. Engineers used prompts such as, “Identify the most influential variables contributing to regenerator temperature instability,” to guide analysis. Scenario simulations helped the engineers evaluate corrective actions and understand downstream implications. Within weeks, temperature variance decreased, emissions excursions ceased and communication across operations, mechanical and process teams improved due to clearer, AI-generated summaries and decision logs. These outcomes reflect the value of AI enabled troubleshooting in complex refinery operations [4].

In an aerospace engineering environment, a mechanical engineer leading a multidisciplinary design team used AI to structure meeting agendas, clarify action items and prepare concise decision summaries. Improved coordination and clearer delegation emerge when engineers use digital workflow tools that structure meeting agendas, highlight action items and generate concise decision summaries for cross-functional teams. Tone-analysis tools helped refine communication strategies and reduce conflict between disciplines. Prompts such as, “Rewrite this agenda to improve clarity and reduce ambiguity,” supported leadership effectiveness. Over several project cycles, meeting duration decreased, delegation improved and design iterations progressed more smoothly as team members gained clarity about roles, expectations and priorities. These results are consistent with research on psychological safety and collaborative engineering [10].

Practical guide for engineers

Engineers can integrate AI into daily practice by incorporating natural-language models into the preparation of technical documents, safety memos and design reviews, ensuring that communication is clear, precise and accessible to diverse audiences. Predictive analytics should be used to support early anomaly detection, strengthen root-cause analysis and provide quantitative justification for operational decisions. AI-generated risk matrices and scenario simulations can enhance hazard and operability (HAZOP) studies and design evaluations by making trade-offs more transparent. Tone-analysis tools can help engineers refine leadership communication, especially when addressing sensitive issues or coordinating across disciplines. AI-based documentation systems can also support audits and knowledge transfer by capturing decision rationales and organizing historical information in a structured, searchable format. These practices align with modern digital work models and productivity research [5, 14].

Key takeaways

The following points summarize the principal insights engineers can apply when incorporating modern digital tools into contemporary engineering practice.

• Engineers can strengthen communication, decision making and collaboration by using modern digital tools that clarify tone, intent and technical meaning

• Structured risk-visualization tools help teams interpret likelihood versus consequence relationships more consistently during abnormal or uncertain operations

• Practical workflows supported by intelligent assistants improve meeting preparation, action-item tracking and cross functional alignment

• Engineers benefit from clearer, more respectful communication when using tone and sentiment analysis to refine how messages are perceived across technical and operational teams

• Real-time decision-support systems help prioritize deviations, highlight emerging risks, and support more proactive operational responses

• Integrating these tools into daily engineering practice enhances both technical performance and essential human-centered skills.

Concluding remarks

Modern engineering practice increasingly depends on the integration of human-centered competencies with advanced digital tools, and this article describes how AI now plays a practical role in strengthening both. Across refining, petrochemicals and energy-transition facilities, engineers are using predictive analytics, natural-language systems, optimization engines and collaboration platforms to enhance communication clarity, improve decision making and support more consistent operational performance. Case studies demonstrate that AI improves safety documentation, accelerates troubleshooting and enables more structured cross-functional coordination, while also reinforcing essential soft skills, such as emotional intelligence, leadership communication and ethical reasoning. As system complexity grows and workforce dynamics evolve, engineers who adopt these tools will be better equipped to manage uncertainty, transfer knowledge and sustain high reliability operations. Integrating AI into daily workflows is not a replacement for engineering judgment, but a means of elevating the human capabilities that underpin safe, efficient and resilient plant performance. 

Edited by Scott Jenkins

References

1. Coker, A. K., “The Soft Skills Edge: Engineering Your Path to Professional Success,” Tellwell Publishing, 2026.

2. National Academy of Engineering, “The Engineer of 2020: Visions of Engineering in the New Century,” National Academies Press, 2004.

3. AIChE, “Competency Framework for Chemical Engineers,” American Institute of Chemical Engineers, 2021.

4. Smith, J., and Garcia, M., AI Applications in Refining Operations: Enhancing Reliability and Decision‑Making, Journal of Process Engineering, Vol. 45, No. 3, pp. 112–128, 2023.

5. McAfee, A., and Brynjolfsson, E., “Machine, Platform, Crowd: Harnessing Our Digital Future,” W. W. Norton & Company, 2017.

6. Goleman, D., “Emotional Intelligence: Why It Can Matter More Than IQ,” Bantam Books, 1995.

7. Felder, R. M. and Brent, R., “Teaching and Learning STEM: A Practical Guide,” Jossey‑Bass, 2016.

8. Kahneman, D., “Thinking, Fast and Slow,” Farrar, Straus and Giroux, 2011.

9. Leveson, N., “Engineering a Safer World: Systems Thinking Applied to Safety,” MIT Press, 2011.

10. Edmondson, A., “The Fearless Organization: Creating Psychological Safety in the Workplace for Learning, Innovation, and Growth,” Wiley, 2019.

11. Shneiderman, B., Human‑Centered Artificial Intelligence: Reliable, Safe & Trustworthy,” International Journal of Human–Computer Interaction, Vol. 36, No. 6, pp. 495–504, 2020.

12. AIChE Center for Chemical Process Safety (CCPS), Process Safety Management Guidelines, AIChE, 2020.

13. Kletz, T., “What Went Wrong? Case Histories of Process Plant Disasters and How They Could Have Been Avoided,” Gulf Professional Publishing, 5th Edition, 2009.

14. Davenport, T. H., & Ronanki, R., Artificial Intelligence for the Real World, Harvard Business Review, Vol. 96, No. 1, pp. 108–116, 2018.

Appendices to this article

Appendix A: AI Prompts and Example AI‑Generated Responses

This appendix provides the full set of prompts referenced in the article, along with representative AI‑generated answers suitable for tools such as Copilot, ChatGPT, or Gemini.

A1. Safety Communication Prompts

Prompt: Rewrite this safety memo for operators with clearer explanation of operating limits and interlocks. Highlight the consequences of exceeding the temperature envelope and propose a simple risk matrix.

AI‑Generated Answer: The normal operating temperature range for the overhead system is 138–145°C. If the temperature exceeds 148°C, the high‑temperature alarm will activate. At 150°C, the high‑temperature trip will automatically close the feed valve and open the reflux‑drum vent to prevent overpressure. Operating above the temperature envelope increases the risk of accelerated fouling in the overhead condenser, loss of separation efficiency, potential overpressure in the reflux drum, and hydrocarbon release if the PSV lifts. The simplified risk matrix indicates medium likelihood, high consequence, and overall high risk, requiring immediate operator action.

A2. Trip‑Logic Clarification Prompt

Prompt: Improve the clarity of this explanation of high‑temperature trip logic.

AI‑Generated Answer: The high‑temperature trip protects the column from thermal runaway. When tray‑10 temperature reaches 150°C, the system closes the feed valve, opens the overhead vent, reduces furnace firing to minimum, and issues a high‑priority alarm. This sequence prevents overpressure and stabilizes the column.

A3. Interlock Explanation Prompt

Prompt: Explain this interlock sequence in simpler language suitable for non‑technical stakeholders.

AI‑Generated Answer: The interlock acts like an automatic safety brake. If the column temperature becomes too high, the system reduces heat, stops new material from entering, and opens a safe vent path to release pressure. This prevents equipment damage and reduces the chance of a hydrocarbon release.

A4. FCCU Troubleshooting Prompts

Prompt: Analyze this FCCU regenerator temperature data and identify possible correlations with air‑blower performance or catalyst circulation. Suggest three plausible root causes.

AI‑Generated Answer: Temperature fluctuations of 8–12°C correlate with a 3–5% drop in air‑blower discharge pressure and a 2–3% reduction in catalyst circulation. Plausible root causes include air‑blower degradation, catalyst slide‑valve sticking, and maldistribution of air due to grid plugging.

Prompt: Simulate the effect of increasing air‑blower capacity by three percent.

AI‑Generated Answer: A 3% increase in air‑blower capacity is predicted to raise dense‑bed temperature by 4–6°C, reduce CO slip by 15–20%, and stabilize temperature fluctuations. Regenerator ΔP may increase slightly due to higher vapor traffic.

A5. Leadership and Collaboration Prompts

Prompt: Summarize this meeting into action items for operations, mechanical, and safety teams.

AI‑Generated Answer: Operations should reduce furnace duty by 5%, monitor reflux‑drum level every 30 minutes, and report deviations. Mechanical should inspect air‑blower filters, verify slide‑valve travel, and schedule condenser cleaning. Safety should update the risk matrix, review alarm‑response procedures, and confirm PSV certification dates.

Prompt: Identify any emotionally charged language in this email and suggest improvements.

AI‑Generated Answer: Emotionally charged phrases include “no one is paying attention” and “operations needs to get its act together.” A more constructive version focuses on system behavior and shared responsibility: “We have observed repeated temperature deviations. Let’s review the procedure together to ensure alignment.”

A6. Agenda‑Clarity Prompt

Prompt: Rewrite this agenda to improve clarity and reduce ambiguity.

AI‑Generated Answer: Revised agenda includes: review of temperature‑control deviations, mechanical status update, FCCU catalyst‑circulation review, safety update on alarm‑response procedures, and confirmation of action items.

A7. Root‑Cause Identification Prompt

Prompt: Identify the most influential variables contributing to regenerator temperature instability.

AI‑Generated Answer: The most influential variables are air‑blower discharge pressure, catalyst circulation rate, CO/CO₂ ratio, dense‑bed ΔP, and feed preheat temperature. These account for approximately 78% of observed temperature variance.

A8. Operator‑Clarity Prompt

Prompt: Rewrite this explanation of the reflux drum high‑level alarm for operator clarity.

AI‑Generated Answer: The high‑level alarm activates at 85% capacity. Continued level rise risks liquid carryover into the overhead compressor. Operators should reduce reflux flow, check cooling‑water performance, and verify overhead‑line pressure.

 

Appendix B: Python Case Study on FCCU Regenerator Temperature Anomaly Detection

Artificial intelligence can support refinery engineers by identifying subtle process deviations that may not be visible through manual review of operating data. The following Python program demonstrates a simple anomaly‑detection workflow applied to FCCU regenerator temperature behavior. The script is fully self‑contained, includes a small built‑in dataset, and can be run directly in Visual Studio or any Python environment. It illustrates how AI tools can help engineers detect abnormal operating patterns, correlate them with air‑blower and catalyst‑circulation performance, and communicate findings more clearly during troubleshooting and decision‑making.

Python Program: FCCU Regenerator Temperature Anomaly Detection

import pandas as pd

from sklearn.ensemble import IsolationForest

# ———————————————————

# FCCU Regenerator Temperature Anomaly Detection

# ———————————————————

print(“\n=== FCCU Regenerator Temperature Anomaly Detection ===\n”)

# Create a built-in FCCU dataset (no external CSV required)

data = {

    “timestamp”: [

        “2024-01-01 00:00”, “2024-01-01 00:01”, “2024-01-01 00:02”,

        “2024-01-01 00:03”, “2024-01-01 00:04”, “2024-01-01 00:05”,

        “2024-01-01 00:06”, “2024-01-01 00:07”, “2024-01-01 00:08”,

        “2024-01-01 00:09” ],

    “regen_temp”: [690, 692, 705, 689, 710, 693, 691, 720, 688, 715],

    “air_blower_dp”: [4.2, 4.1, 3.8, 4.3, 3.7, 4.2, 4.1, 3.6, 4.3, 3.7],

    “catalyst_circ_rate”: [28.5, 28.4, 27.9, 28.6, 27.8, 28.5, 28.4, 27.7, 28.6, 27.8]

}

 

df = pd.DataFrame(data)

 

print(“Built-in dataset created.”)

print(“Columns:”, df.columns.tolist())

print(“Total data points:”, len(df))

 

# Select key variables for analysis

X = df[[“regen_temp”, “air_blower_dp”, “catalyst_circ_rate”]]

 

# Fit anomaly detection model

model = IsolationForest(contamination=0.15, random_state=42)

df[“anomaly”] = model.fit_predict(X)

 

# Extract abnormal events

abnormal = df[df[“anomaly”] == -1]

 

print(“\nNumber of abnormal events detected:”, len(abnormal))

print(“\nSample of abnormal readings:”)

print(abnormal[[“timestamp”, “regen_temp”, “air_blower_dp”, “catalyst_circ_rate”]])

 

# Summary statistics

normal_avg = df[df[“anomaly”] == 1][“regen_temp”].mean()

abnormal_avg = abnormal[“regen_temp”].mean()

 

print(“\nAverage normal regenerator temperature:”, round(normal_avg, 2), “°C”)

print(“Average abnormal regenerator temperature:”, round(abnormal_avg, 2), “°C”)

 

print(“\nAnalysis complete.”)

print(“Abnormal events typically indicate air-blower instability or catalyst circulation issues.\n”)

Results

[Running] python -u “c:\Users\Kayode Coker\OneDrive\Desktop\Python_Folder\FCCU Regenerator_Temperature_Anomaly_Detection_akc1.py”

=== FCCU Regenerator Temperature Anomaly Detection ===

Built-in dataset created.

Columns: [‘timestamp’, ‘regen_temp’, ‘air_blower_dp’, ‘catalyst_circ_rate’]

Total data points: 10

Number of abnormal events detected: 2

Sample of abnormal readings:

          timestamp  regen_temp  air_blower_dp  catalyst_circ_rate

2  2024-01-01 00:02         705            3.8                27.9

7  2024-01-01 00:07         720            3.6                27.7

Average normal regenerator temperature: 696.0ºC

Average abnormal regenerator temperature: 712.5ºC

Analysis complete.

Abnormal events typically indicate air-blower instability or catalyst circulation issues.

[Done] exited with code=0 in 1.474 seconds

Interpretation

When executed, the program identifies abnormal regenerator‑temperature events based on multivariate patterns involving air‑blower differential pressure and catalyst‑circulation rate. These anomalies often correspond to early indicators of combustion instability, maldistribution, or mechanical degradation. The output provides engineers with a concise summary of abnormal time stamps, temperature deviations, and comparative averages, supporting more structured troubleshooting and clearer communication with operations and maintenance teams.

Acknowledgement

The images appearing in this article were generated by the author.

Author

A. Kayode Coker, (kcoker1@hotmail.com) is an engineering consultant with AKC Technology (www.akctechnology.com) and an Honorary Research Fellow at the University of Wolverhampton, U.K. He previously served as Engineering Coordinator at the Saudi Aramco Shell Refinery Company and as chair of the Chemical Engineering Technology Department at Jubail Industrial College. A chartered chemical engineer for more than 40 years, he is a Fellow of the Institution of Chemical Engineers and a senior member of AIChE.Coker holds B.Sc., M.Sc. and Ph.D. degrees in chemical engineering from Aston University, U.K., and a teaching qualification from the University of London. He has taught extensively, delivered international short courses, and published widely in chemical and petroleum engineering. He is the author of 14 books, a contributor to the Encyclopedia of Chemical Processing and Design, and a certified mentor trainer and technical assessor for IChemE.