The hypothesis
that diversity among teams increases effectiveness is well-researched and published.1
While much of this research focuses on strategic leadership teams, the same
principles applied at the tactical level in process manufacturing frequently
foster new and innovative approaches to problem-solving at the frontline.
Reliability,
machinery, fixed-equipment, electrical/instrumentation/controls, and
maintenance teams tend to have their own ways of thinking and working, focused
on design conditions and vendor recommendations. Though their work processes
intersect on an as-needed basis, many tasks tend to be executed in silos.
Placing
a process engineering lens on what is typically considered mechanical or
electrical engineering challenges can transfer learnings and best practices
across the different technical functions in an organization. Advanced analytics
and other statistical methods typically applied to product and process problems
have an understated utility in maintenance and reliability applications.
Barriers
to cross-functional connectedness. Creating
an effective modern reliability and maintenance engineering team starts with ensuring
members can access all the data needed to predict failures, identify
root causes and optimize condition-based maintenance strategies. Historically,
this data set included maintenance databases, high-frequency machinery
monitoring information, piping and instrumentation diagrams (P&IDs), and
electrical and control loop drawings. However, often missing from these
analyses was the data demonstrating what is going on in the process right now.
This piece is critical because it enables a comparison between present operations
and those of several years ago.
When plant
personnel notice operational issues, they typically engage a machinery engineer,
who requests data from a process engineer, which is usually shared via an offline
spreadsheet. This ignites a communication flywheel rich in dependencies but
lacking in collaboration. Desktop-based spreadsheets have a long tenure as the
primary tool for analyzing data from multiple sources because of their
availability and familiarity. However, spreadsheet-based analyses heighten the existing
barriers to effective collaboration for solving maintenance and reliability
challenges.
In addition
to stifling collaboration—which silos efforts in initial solution development—the
absence of live data connectivity makes many analytics platforms unfit for near-real-time
analytics, such as monitoring, predictive and prescriptive. Once an issue’s root
cause is identified, monitoring measures must be put in place to prevent
similar failure modes from occurring in the future.
The
calculations built into most analytics software solutions require input data
feeds from multiple source systems. However, without live connections to all
relevant data feeds, monitoring efforts are left to manual means, with
calculations refreshed only when new data sets are queried and wrangled into
the offline solution.
One access point, endless capabilities. Modern,
cloud-based advanced analytics solutions make it easy for entire organizations
to access data from hundreds of different data sources through a single pane of
glass. With process, maintenance, reliability and other engineering functions
all accessing the data they need via the same application, sharing is innate,
and the many email chains requesting, gathering and assembling data become relics
of the past. Browser-based—rather than desktop-based—applications, in
particular, facilitate real-time collaboration with device-agnostic
compatibility.
At
the same time, a process engineer is looking for anomalies amongst historian
sensor data, a machinery engineer can overlay recent high-frequency equipment
data, while a maintenance engineer adds event data from historical workorders
for context. As a team, they build a more efficient, informative and
multi-perspective workflow, empowering deeper understanding of the variables
and effects at play than individualized troubleshooting workflows would have
yielded.
For
organizational buy-in to hold, collaboration capabilities must be realized quickly.
Advanced analytics software is architected to connect rather than copy
and move data, ensuring existing data infrastructure remains the authoritative
single source of truth. This mechanism can also reduce the time to value for
software implementation from months or years—typical of a large-scale data lake
migration project—to hours or days. The rapid return on investment of modern
advanced analytics software is a selling point to operations and information
technology teams, many of whom are striving to achieve returns on multimillion-dollar
digital transformation initiatives.
Accessibility and collaboration produce
novel industry solutions. Throughout
the petrochemical and refining industries, advanced analytics solutions are
fostering collaboration between reliability, maintenance and process teams. The
results of these collaborations are unique, sustainable monitoring solutions
for maintenance processes that were historically reactive. The following case
studies demonstrate the value that can be attained by implementing these types
of solutions.
Golden
batch as an outcome of root cause analysis. When a critical feed gas
compressor on a polyethylene line at a petrochemical plant tripped and was
unavailable for immediate restart, a reactor shutdown ensued. An unplanned
reactor shutdown of this sort at the facility typically lasts a minimum of 4 hr,
costing the plant upwards of $200,000. The compressors were maintained on a
preventative maintenance schedule recommended by the manufacturer, but despite
strict adherence to the robust schedule, unplanned shutdowns still occasionally
occured.
Immediately
following the compressor trip, machinery engineers reported to the scene,
combing through high-frequency data. However, due to software storage
constraints, this information only dated back 30 d. The controls team
identified the safety interlock that prompted the shutdown, ushering in electrical
engineers to the investigation team. This group traced electrical diagrams all around
the pump motor, which was time-consuming and ultimately did not reveal a root
cause.
Meanwhile,
a process engineer, armed with decades of historical data in an advanced
analytics application, quickly located the five most recent shutdowns. Using
technology in the software called “capsules,” the engineer was able to focus
the investigation on these shutdown and subsequent start-up time periods. By leveraging
an augmented time visualizationa, the engineer examined the startups
following each of the recent shutdowns simultaneously, and noticed an abnormal
discharge pressure profile in the two most recent startups (FIG. 1).
Investigating further, the engineer also noticed early warning signs on the
motor amperage signal. Without a way to view the start-ups overlaid, the motor
degradation had gone unnoticed by operations.
As a
result of the root cause analysis, the process engineer was tasked with putting
a monitoring solution in place to provide motor degradation insight and prevent
a similar unplanned shutdown in the future. The manufacturer used the author’s
company’s advanced analytics solution to create logically defined capsules for
each shutdown and subsequent startup of the compressor. Referencing ampere and
pressure data from a few healthy startups, golden profiles were created to
monitor each signal, and flag when sensor values were outside a healthy range.
Today,
when an out-of-range value first appears, the compressor motor is added to the
maintenance work list for the next planned shutdown. This proactive maintenance
approach, resulting from golden batch-style monitoring, is expected to
eliminate unplanned shutdowns due to this failure mode.
Capability
analysis for controller performance monitoring. Process
capability analysis, calculated as Cpk, is a commonly used statistical tool for
measuring a production unit's ability to produce a product at target and within
its quality specification limits. While the quality parameter application of
Cpk is a longstanding component of the Six-Sigma methodology, its application
in control loop analysis remains relatively untapped. Control valves are sized—typically
during initial project engineering and design phases—to deliver flowrates
within an expected “good controllability” range of output percentages (OP%). As
product grade slates evolve in response to customer demand, variable setpoint
targets shift and can wander far from the center of the control valve's
controllable range.
A
specialty elastomer manufacturer was struggling to control a critical feed
component flow rate at setpoint. Meeting the desired setpoint required the
controller to adjust the OP% value well below the originally designed range,
resulting in oscillations. These oscillations propagated through the system and
caused product composition swings, resulting in quality downgrades.
Process
engineers calculated the Cpk of the controller using the valve’s design output
range, and they found the Cpk had been declining for the past decade. Senior
process operators hypothesized the decline seemed to coincide with the unit
changing product grade slate, which the data confirmed. For this reason, the
control valve was resized, and the issue subsided.
With
the dramatic product grade slate shift, plant management was concerned other
controllers were at risk of causing similar quality issues. To address this
concern, engineers performed a plantwide Cpk analysis on all product
composition critical control valves. Each valve's process value, setpoint and
output signals were structured into an asset group, along with the valve's high
and low output design limits. This enabled engineers to create the Cpk
calculation for one valve, then immediately scale it across all composition
critical control valves.
Roll-up
visualizations (FIG. 2)
provided the team a live look at the current Cpk of each valve. Then, they
calculated the increase in decline rate of Cpk values over the past decade and
created a table prioritizing valves with the sharpest decline. A monitoring
solution was put in place for proactive detection and replacement of improperly
sized control valves going forward.
Inclusive teams develop innovative
solutions. The collaborative nature of
modern advanced analytics solutions encourages distinct engineering functions—which
have traditionally been siloed—to come together and solve problems faster. Assembling
teams of diverse technical backgrounds increases the rate at which problems are
addressed, and it expands the set of plausible solutions beyond the status quo.
This fosters innovative and sustainable solutions to maintenance and
reliability challenges capable of standing the tests of time and technology. HP
NOTES
a
Seeq Corp.’s Chain View
LITERATURE CITED
ALLISON BUENEMANN is the Chemicals Industry Principal at Seeq Corporation. She has a process engineering background with a BS degree in chemical engineering from Purdue University and an MBA from Louisiana State University. Buenemann has nearly a decade of experience working for and with bulk and specialty chemical manufacturers like ExxonMobil Chemical and Eastman to solve high-value business problems leveraging time series data. In her current role, she enjoys monitoring the rapidly changing trends surrounding digital transformation in the chemical industry and translating them into product requirements for Seeq.