What is Process Health?
There's already some discussion started on Machine Health and Production Health, but I'm interested if there is a common understanding of Process Health, what constitutes Process Health, how it can be assessed and the overall value of focusing on this.
Comments
-
Hey Scott! Great question. Process Health uses artificial intelligence (AI) to prevent inefficiencies and production losses such as quality, yield, throughput, waste and energy. It allows manufacturers to realize the full potential of production lines, while reducing months of complex engineering work to minutes.
Hope that helps! 😊
1 -
Great callout on the categories. The ability to characterize those as imputs AND outputs against other manufacturing metrics or KPI's, as well as the benchmarking of these elements is what makes the process health category so powerful and difficult to ignore as an opportunity for not only OEE and Utilization considerations but Production Health overall.
0 -
As mentioned, my perspective is that Process Health relates to the overall process inputs, outputs, OEE, efficiency, energy, etc....some of which incorporate machine health to make sure equipment is running as intended. Along with using Augury for a portion of our machine health, we are also in the middle of a 1-year pilot with Seebo (recently acquired by Augury) to utilize multi-variable statistical process control via machine learning AI to characterize and recommended process conditions for two of our production units (rotary dryers).
I know it's Augury's goal to integrate the Seebo process health with Augury machine health platforms (hence the acquisition), but I honestly struggle to see how that will happen.
3 -
There's no right or wrong...I tend to capture the manufacturing metrics/KPI's under Production Health. Process Health in my world were all the process instrumentation, process variables (measured and/or calculated), auto and even manual machine settings (as input into and captured from a digital log, HMI, Supervisory Control System), Operator ID's, environmental data/sensors, PID control changes and output/output saturation amongst others. I have also stood up a seperate category called Material Health where CofA's of all raw materials are captured along with any routine sampling/inspection and test data and any other data related to RM attributes (moisture content, size distribution, etc.). I think in the Seebo world this is all included in Process Health. Again, no right or wrong. I see Production Health as the aggregate of all healths and what's may be most related to servicing the customer.
At GAF Production Health = Machine Health + Process Health + Material Health + Product Health (conformance to spect, bill of materials efficiency, ratio to standard cost, amongst others). So it may be easier to think of Machine Health and Process Health as complimentary and related. Definitely there are many machine issues that create process upsets, whether direct (operator response) or indirect (control system/output response). Knowing what creates what is the bigger part of the process excellence battle.
1 -
Thanks @Scott Reed. Good assessment. I like your equation...At GAF Production Health = Machine Health + Process Health + Material Health + Product Health and you've added both RM and finished product quality to the discussion.
1 -
@Kevin Loos, great question about how Machine Health would integrate with Process Health. We’re taking steps to integrate around specific use cases.
As @Scott Reed mentioned, machine conditions often influence process outcomes, and in some cases process parameters can lead to unexpected machine failure.
Our goal is to start with customer use cases where insights into the relationship between machine and process health are particularly impactful, and where correlating maintenance and operational activities would result in significant efficiency improvements.
Here are a few use cases:
- Optimizing maintenance using operational insights. For example, many of our food processing customers overhaul extruders preemptively, as clogging can cause signing any quality issues and downtime. Overhauls are expensive, but without the ability to identify clogging ahead of time, the extruders are over-maintained on a frequent schedule. Clogging cannot be effectively detected using mechanical data, but we will be able to identify clogging patterns using process data. Once clogging is detected, we could provide a prescriptive recommendation for optimal maintenance timing and potentially process set points that would keep the extruder running in the meantime.
- Optimizing process using mechanical insights. For example, we've helped identify the mechanical root cause for quality issues on a high-speed CPG production line. The site team spent months trying to understand the the reason for an intermittent quality issue that seemed random. After deploying machine health, we combined mechanical data with an operational understanding to help uncover a failing gearbox component that was used only for a specific size of product. The customer was able to adjust the product mix, and change the component without further affecting production goals.
- Optimizing an entire production system. For example, the efficiency of steam systems is highly dependent on both the process of steam generation and proper maintenance of steam distribution. We've helped an Oil & Gas customer optimize steam generation with process health, and are working with the same customer on predictive maintenance of pumps and steam traps across their large steam distribution system. As a result, we expect a dramatic reduction of lost steam, and higher overall energy efficiency and sustainability.
2 -
Excellent examples Artem provides as use cases - there are infinitely more. In web handling processes MD and CD weight and thickness consistency is paramount and there are direct relationships in achieving precision between the process itself (chemistry, materials, physical properties) and the machine components delivering the materials throughout the manufacturing line. So much of the PdM is driven by the Process Health conditions in these operations.
In each of the health categories, I tend to think in terms of the Plan, Source, Make & Deliver domains and then further break the domains down to the Ishikawa 6 blocks - man, machine, methods, measures, materials, environment. Every Health may not have the Deliver or Source domains represented, and certainly every domain may not be impacted by environmental concerns, but for me this is a good way to understand how broadly the manufacturing may be characterized in terms of the opportunities to achieve operational excellence and maximize customer experience.
0 -
Hi @Kevin Loos - thanks for your question and trust that @Artem Kroupenev and @Scott Reed responses provided some useful context. If you're interested in going deeper, we'd be happy to set up a convenient time to engage.
1 -
Great examples provided by @Artem Kroupenev. In my experience, combining PH and MH have significant in efficiency and quality as mentioned, but also in customer service, safety and cash flow. And helps to improve and simplify training and operation in the plant floor.
2 -
Hi @Chris Dobbrow I was involved/sponsored Seebo integrations and unfortunately share the same struggle described by @Kevin Loos.
1 -
@Kevin Loos &@Omri Zurawel, can either of your provide more context as to the "struggle" to understand how Augury will integrate the Seebo process health with Augury machine health platforms. Is this more related to Augury support structure or how the healths work in relation to one another?
0 -
Just wanted to kick this up to the top of the "recently commented" sort menu. If there is still a lack of understanding of how Process Health can incorporate Machine Health within Augury/Seebo, I'd sure like to appreciate this concern better. The journey to Prescriptive control must include the insights from a Process Health solution, IMHO.
0
