Paper360 - March/April 2018 - 30

data systems. Typically mills have tens of
thousands of process measurements and
control loops where data are collected.
Traditional data systems and methods have
only allowed for the analysis of a very small
fragment of the available data. Data analysis has been time consuming and laborious.
Today's more advanced tools provide easy
analysis of large data sets and propose relevant, expedited answers to complicated
questions. As a result, mills are able to find
valuable optimization potential hidden in
process data, bringing results to everyday
operations and decision making.
Decisions that tackle acute problems or
mill process adjustments need to be made
quickly; valid, accurate answers can now be
realized immediately. Plus, efficient data
systems can ignore irrelevant data, which
in the past would normally destroy analyzing attempts. Innovative data systems can
propose trusted answers to questions such
as the root cause of unwanted process events
and allow for fast fail testing hypothesis in
order to eliminate incorrect assumptions.
WHAT IS THE BIGGEST MISTAKE MILLS
MAKE WHEN IMPLEMENTING A NEW
DATA SYSTEM?
Vikman: In any engineering project, the
plant data management system and implementation work must start early, so that it is ready
when the plant or piece of infrastructure starts
operation. For this, all user needs-such as
preventive maintenance plans, material master integration to purchasing systems, and
material storage work-must be started early
enough. The standards need to be in place and
clear to every supplier as they provide their
technical data to be uploaded in the plant data
management system.
Gee: A data system itself does not improve
mill performance. The data system must be
well designed to integrate easily with each
mills' everyday operations and management
goals to achieve best results. This means:
* The same technology platform should be
used throughout the organization to support mill goals. This gives management
visibility to process and operations; engineering a focus on improvement potential
and solving bottlenecks; and operators a
user-friendly tool to follow best practices
for continuous improvement.
* Data access is often limited to only a subset of data, resulting in a low resolution
compared to process dynamics; mill data
30

Paper360º

MARCH/APRIL 2018

systems must provide real-time, on-line
access to all process data.
* In the past, data systems chosen have not
been easy for users to use. The results are
poor utilization limited to a few specialists,
resulting in increased risk.
Westerberg: As a data systems provider,
the biggest mistake we see mills make is not
enough effort is spent early on in the project
cycle to fully understand the user's needs.
Later in the project this results in excessive
customization on the system to suit the needs
of the operation. In the end the project comes
in way over budget, late, and has tremendous
issues to support long term.
The best approach is to keep to the 80/20
rule: about 80 percent should be standard
functionality, with at most 20 percent customization, for a particular project.
WHAT'S THE MOST IMPORTANT THING
A MILL CAN DO TO GET THE MOST
FROM ITS DATA SYSTEM?
Gee: Mills need to do their due diligence-
select a data system with the necessary power,
intelligence, and ease of use to deliver results
and provide seamless integration with everyday work processes and management practices. This includes:
* Select a proven data system that is easy
to use for all levels of the organization,
efficient in a mill environment, and
demonstrates benefit with real world
process data (outlier values, shutdowns,
grade dependency.)
* Integrate seamlessly with everyday operations and management systems.
* Don't make an assumption before the
analysis is performed; i.e., don't try to find
the process data to support a conclusion or
assumption. Determine the question(s),
use all possible data for analysis and diagnostics, and draw a conclusion based on
the data. This requires all data to be available on-line at all times and will set the
requirements needed to support the mill
data system.
* Data system should be flexible to accommodate new events without configuration.
Westerberg: The system must be designed
to be easy for a user to get the information
they need and to share the results with others in the organization without having to be
an expert in the underlying system technology. Information must be accessible from any
location, whether it's in the mill, at home or
at a conference, 24/7, and from any device.

Vikman: To start a plant data management system project, the first thing is to
define the standards (technical and commercial), classifications and equipment, and
department hierarchy that will be the base
for the material master. The standards must
be chosen based on the industry regulations
and local deviations (ISO, TÜV, ASME, PSK,
etc.) Also, the network architecture and data
communications rules need to be defined at
the beginning.
The system will be as good as the users'
ability to take advantage of the possibilities
it creates. The basic needs must be identified
before the system is selected and decisions
made on how the system can be integrated
with other data sources, e.g., ERP-system.
The possibilities are huge, starting from basic
document archive to real 3D mill with real
information about the whole site.
WHAT'S THE NEWEST WAY YOUR
COMPANY IS ADDRESSING MILL
DATA NEEDS?
Westerberg: To address the need for
enterprise wide data history and analysis, Honeywell recently introduced the
"Uniformance Connected Historian" cloudbased Big Data enterprise historian. It is
offered as a cloud subscription instead of the
traditional large capex and resource-intensive
IT project, which significantly reduces total
cost of ownership for our customers.
Vikman: Pöyry has years of experience
of plants' data management systems, starting from pre- and basic engineering phases
and covering the whole lifecycle of the plant.
Pöyry Virtual Site design tool information
covers all engineering disciplines and can
be integrated with all relevant sources and
enriched, e.g., with machine supplier 3D
models or aerial photogrammetry of site.
Gee: Trimble's Wedge is an online process
diagnostics system that manages Big Data
from different sources for expedited analysis
and diagnosis of the root cause of a problem.
Wedge's easy user experience gets process
operators and engineers up and running
instantly without the need for developers and
programmers. Wedge delivers an auditable
version of the truth and is the predominant
solution to routinely tune an organization's
process performance and influence operational efficiencies. WestRock currently uses
Wedge as a PI data analysis tool to find data
relationships that would not have been visible
without it.
www.tappi.org

http://www.tappi.org

Table of Contents for the Digital Edition of Paper360 - March/April 2018