Keith Smith

Siloed data is killing your factory’s speed. Break the walls or get left behind. Let’s be real. Most factories still run with data trapped in separate systems. Machines talk to PLCs. Operators fill out paper logs. MES and ERP live in their own little worlds. What does that mean for you? → Slow decisions. → Missed signals. → Unplanned downtime. → Quality issues that slip through the cracks. I see it almost every week—manual quality checks, rule-based automation, and teams reacting after problems hit. Scheduling feels like a guessing game. Asset optimization? More gut feel than data-driven. But here’s where things get exciting: Agentic Manufacturing Operating Systems. These are not buzzwords. They’re a real shift, and we're building it at iosys: → Machines observe, plan, act, and learn. → Systems adapt to what’s happening right now, not yesterday. → Data flows across the shopfloor, not stuck in silos. What can you unlock? - Predictive maintenance. Eliminate unplanned breakdowns. - Adaptive scheduling. Allocate resources where efficiency is highest. - Automated quality control. End reliance on sample checks. - Intelligent inventory management. Reduce waste, maximize profitability. - Seamless human-machine collaboration. Enhance performance across operations. I’ve seen the change. When data connects, factories move from “black box” to “glass box.” Suddenly, you spot patterns. You solve problems before they grow. You win back speed, flexibility, and trust. Enabling people and processes isn’t about fancy tech for the sake of it. It’s about clear, shared data that powers real action. Are you still letting siloed data hold you back, or are you ready to break the walls? What’s your experience with data integration on the shopfloor?

9

2 Comments

Responsive visualizations often need code, but does it always need to run? One of the strengths of Codesys visualization is how tightly it integrates with the PLC logic, offering huge flexibility for custom implementations (if you take the time to hunt down the nuggets of info floating around). Take the analog indicator shown here. It needs scaling logic to normalize values for the ISA-style indicator. Initially, I placed this logic in its own FB inside my Analog Input FB. That meant passing variables in and out, and calling the FB continuously, even if the ISA bar wasn’t being used. Eventually I refactored it into a method inside the same FB. Cleaner, fewer variables, but still being called every cycle. Then Jeff Nimtz showed me another way. You can call the method directly from the visualization, and not via an action or event like you might expect. In Codesys, it's possible to call a method from one of the visualization object's variable properties. In my case, I linked the method to the text variable property of the background rectangle. The result? The method is called only while the object is visible on screen, and it stops when the object is no longer active. No unnecessary computation, no wasted cycles. Performance is cheap these days and readability is a top priority in PLCs, so do the performance gains necessitate the slight abstraction? Let me know your thoughts below. #codesys #twincat #plcprogramming #ladderlogic #structuredtext #systemintegrator #agautomation

64

14 Comments

Why Modernization Doesn't Mean "Rip and Replace" Phased Automation Upgrades: What's Actually Possible One of the biggest misconceptions about upgrading obsolete automation systems: "We'd have to shut down for weeks and replace everything." Not true. Here's what phased modernization actually looks like: Assessment Phase (1-2 weeks): Document existing system architecture Identify highest-risk components Evaluate what can be retained vs. what needs replacement Determine integration points with existing equipment Planning Phase (2-4 weeks): Design new control architecture Program and test new systems offline Create cutover procedures Schedule around production demands Implementation Phase (varies): Often done during scheduled shutdowns or low-production periods New and old systems can run in parallel during transition Existing field devices (sensors, valves, motors) typically remain Wiring infrastructure often stays in place Real Example: A 26-year-old PLC system controlling multiple production lines was converted to Siemens S7-1500 over 3 months. Total production downtime? 4 days spread across 3 planned maintenance weekends. What typically stays: Motor control centers Field instrumentation Pneumatic systems Physical I/O wiring (often) Existing HMI screens (upgraded software) What gets replaced: PLCs and processors Communication networks Control logic (rewritten but functionally identical) Outdated or failed components The goal: Minimize disruption while eliminating obsolescence risk. It's possible—we do it regularly. Have questions about what a conversion would look like for your specific system? The assessment phase answers exactly that.

6

3 Comments

𝐖𝐢𝐝𝐞 𝐯𝐬 𝐍𝐚𝐫𝐫𝐨𝐰 𝐓𝐚𝐛𝐥𝐞𝐬 𝐢𝐧 𝐓𝐢𝐦𝐞 𝐒𝐞𝐫𝐢𝐞𝐬 𝐃𝐚𝐭𝐚𝐛𝐚𝐬𝐞𝐬 One of the big benefits of new time series databases in manufacturing is that you can define your own schema. Databases like QuestDB, TimescaleDB, InfluxDB, and AVEVA Connect offer that capability. So it's not just timestamp, value, and maybe a quality flag—you can add more columns and choose from a larger number of data types. This enables new features: 𝐋𝐚𝐭𝐞𝐧𝐜𝐲: One of the main sources of uncertainty when working with time series is that you have a source timestamp for the data point, but that doesn't mean this is when the point arrives in the historian or analytics system. Often, this arrival time is a major culprit when triggering events or dependency calculations. Capturing the arrival timestamp would allow for easy analysis of these edge cases. 𝐒𝐞𝐪𝐮𝐞𝐧𝐭𝐢𝐚𝐥 𝐈𝐃: Manufacturing data flows through many subsystems before being stored: PLC, OPC-UA, interface, broker, and more. How do you know if a message package isn't dropped somewhere along the way? Tagging each point with a sequential ID would give you more confidence in your data quality. 𝐇𝐚𝐬𝐡𝐢𝐧𝐠: Creating a verification hash on the payload would make your entire data pipeline tamper-proof. This might be a feature that regulated industries would find valuable to button up security. But you can also use additional columns to create wider tables per tag or stream if the data is evenly sampled. Below is an example from the TimescaleDB website. These wider tables have advantages over their narrow analogues, such as faster querying of related information and smaller storage footprints since you're not repeating the timestamp for each metric. They also simplify correlation analysis between signals captured at the same moment. However, wide tables come with trade-offs: they work best when all columns are sampled at the same rate, they can become sparse if sensors report at different intervals, and schema changes require more planning. Choosing to implement wide tables, or mixing formats, shifts complexity to client applications. Developers must handle multiple query patterns, pivot or unpivot data depending on the source, and reconcile different column structures when joining data. What was once a uniform API call becomes table-specific logic, increasing maintenance burden and reducing interoperability with tools expecting a standard narrow format. What approach are you using, 🔹wide tables with multiple metrics per row, 🔹or narrow tables with a tag-value structure, 🔹a blend, and have you run into limitations with either design?

27

5 Comments

Legacy System Migration in 3 Weeks: Where Seniority Makes the Difference. There comes a moment when every Plant Manager must face the inevitable: core automation is becoming an operational risk. The system is obsolete, documentation is fragmented, and the fear of prolonged downtime stalls any modernization initiative. This is where my role begins: where standard solutions end. I recently managed a critical challenge in the aluminum extrusion sector: The Objective: Migrate the entire core automation from legacy systems to a modern Industrial Ethernet architecture. The Constraint: A shutdown window of only 3 weeks. The Challenge: Fragmented software across old versions, without a solid documentation base. How do you win such a challenge? Knowing the code isn't enough. It requires intuition built over nearly 40 years in the field. I performed surgical reverse engineering, unifying complex code to ensure every process logic was preserved and optimized. The Result: The plant resumed full production exactly on schedule. Throughout my career, I’ve learned that true seniority transforms a catastrophic risk into a high-performance asset. To learn more about this project and explore other international case studies, I invite you to check the Experience section of my profile. If your company faces complex challenges requiring uncommon seniority, let’s connect. #IndustrialAutomation #LegacyMigration #RockwellAutomation #SeniorEngineering #ProblemSolving #Industry40 #AerreSolution

2

Most energy reports end up in a folder. Nobody acts on them. So, I stopped making reports. I built an operating system instead. ───────────────────────── Introducing the Energy Efficiency Command Center ───────────────────────── One Excel workbook. 20 sheets. Multiple buildings across multiple cities. 2 years of data. Zero manual effort to refresh. Here's what it does — automatically: 🏢 PORTFOLIO OVERVIEW Total consumption, cost, CO₂, EUI, and year-on-year change across all buildings at a glance. 📊 RANKINGS + BENCHMARKING Every building ranked by Energy Use Intensity (kWh/m²/yr) against ASHRAE 90.1-2019 baselines. You see in 3 seconds which buildings need urgent action and which are performing. 📈 TREND ANALYSIS Monthly consumption curves overlaid with cooling degree days. So, when someone asks, "why did August cost more?" — the chart already answers it. 💰 COST DECONSTRUCTION Energy charge vs demand charge vs fixed charge — stacked visually every month. Because demand charges are often the silent killer in Saudi SEC bills and most teams don't even track them separately. 💡 OPPORTUNITY FINDER Rule-based engine that flags: high baseload, after-hours waste, low load factor, demand spikes, weather-normalized underperformance. Each flag comes with a SAR savings estimate (low/mid/high band). The portfolio is sitting on ~1.6M SAR/yr in identified savings. 🏅 STANDARDS COMPLIANCE — BUILT IN Not just EUI numbers. The tool calculates: → ASHRAE 90.1-2019: % improvement vs baseline, per building → ASHRAE 100-2018: Performance against existing building targets → LEED BD+C v4.1: Indicative EA points (max 20) → LEED O+M v4.1: Existing building energy performance score → EDGE v3: Energy reduction % toward certification thresholds All of this auto-populates from the same utility bill data. 🖨️ BUILDING REPORTS Select any building + year → one A4 printable page with full KPI summary, monthly trend, opportunity flags, and standards compliance snapshot. Ready for management review in 30 seconds. ───────────────────────── What this is really about: ───────────────────────── FM teams in the GCC region are sitting on enormous energy waste and in some portfolios, 20-40% above ASHRAE baselines. The problem isn't lack of data. It's the gap between data and action. This tool is that bridge: Data → Decision → Action → Evidence → Management Visibility That loop, running continuously, is what transforms an FM operation from reactive maintenance to proactive energy control. Building this kind of infrastructure is what I do. If your portfolio doesn't have a system like this yet, it should. ────────── 💬 Happy to share the methodology or discuss how something like this could work for your portfolio. DM open. #EnergyEfficiency #FacilitiesManagement #Sustainability #GreenBuilding #LEED #EDGE #ASHRAE #SaudiArabia #Vision2030 #SmartBuildings #EnergyManagement #FM #BIMandFM #DataDriven #NetZero #BuildingPerformance #Excel #PowerPlatform #DigitalTransformation

30

3 Comments

Thought for the day: Do today's enterprise dashboards parallel plant alarms of the '80's, '90's, and early 2000's? The other week I made a quick post asking if alarms were obsolete and got a bunch of great responses. Many of them talked about how many alarms now led to automatic actions and how alarm management standards had been implemented to help deal with over-alarming. When we look at these solutions and others, they may seem obvious, but they were developed due to situations where alarm floods, nuisance alarms, and other issues stemming from poor alarm management caused real, sometimes fatal, damage. While these issues may have existed for a while, they became a much bigger one with plant-wide SCADA and the ability to take tons of information and put it in front of an operator. With many technological advances, the technology can outpace the processes needed to deal with them. Plant communications did this before with alarms. Are enterprise communications now doing this with dashboards? We have the ability to gather and project tons of information in front of business and operations leaders, but are we lacking the standards, processes, and automation needed to effectively handle this information? Will those responsible for monitoring this information miss signals due to information overload, ignore areas due to poor data constantly causing poor signals, or not know what to do with the data when they get it? Would like to know what you think, especially those that commented on the first post: Dylan DuFresne, Vladimir Romanov, Mark Firmin, Sr., John A. Whitten, Federico Walas Mateo, Bryce Yokom #alarms #IACS #dashboards

6

9 Comments

𝗢𝗻𝗲 𝘁𝗲𝘀𝘁 𝘁𝗵𝗮𝘁 𝗳𝗶𝗹𝘁𝗲𝗿𝘀 𝗼𝘂𝘁 𝗲𝗺𝗽𝘁𝘆 𝗮𝘂𝘁𝗵𝗼𝗿𝗶𝘁𝘆 Here’s a simple test most public writing fails in manufacturing: Would you be comfortable rereading this after a project post-mortem? In this industry, credibility is stress-tested. Assumptions get revisited. Decisions get questioned. Outcomes get examined with hindsight and numbers. Writing that avoids pressure may feel safe in the moment, but it doesn’t signal experience to anyone who has lived through that process. Authority isn’t built by saying less. It’s built by showing that your reasoning can survive scrutiny when conditions change. That doesn’t require drama. It requires clarity about constraints, tradeoffs, and responsibility. Anything else is noise. 𝗣.𝗦. 𝗜𝗳 𝗮 𝗽𝗼𝘀𝘁 𝗰𝗮𝗻’𝘁 𝘀𝘂𝗿𝘃𝗶𝘃𝗲 𝗿𝗲𝘃𝗶𝗲𝘄, 𝗶𝘁 𝘄𝗼𝗻’𝘁 𝘀𝘂𝗿𝘃𝗶𝘃𝗲 𝘁𝗵𝗲 𝗺𝗮𝗿𝗸𝗲𝘁 𝗲𝗶𝘁𝗵𝗲𝗿. 𝗧𝗵𝗮𝘁’𝘀 𝘁𝗵𝗲 𝘀𝘁𝗮𝗻𝗱𝗮𝗿𝗱 𝗺𝗼𝘀𝘁 𝗶𝗻𝗱𝗶𝘃𝗶𝗱𝘂𝗮𝗹𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗿𝗲𝗰𝗼𝗴𝗻𝗶𝘇𝗲.

2