Paper Mill Optimization Savings Add Up. One of the three machines at papermaker’s South Carolina facility wasn’t able to transition quickly enough between products. As a continuous process, this meant mountains of off-spec material was produced during the transition. ABB Optimization Services experts relied on the Transition Fingerprint process to identify both hardware and software issues. Following optimization, transition time was reduced from 33.4 to 23.3 minutes, generating material, energy and other savings of $387K/year.
Conduct a Test Plan
The data may reveal issues in a variety of places. It could be found in:
- Process instrumentation, the devices that measure production parameters
- Process controllers, the software that looks at the instrumentation, compares it to a target, and calculates the desired actuator setpoint change
- Process actuator, the motors, pumps, valves or other mechanical devices.
Based on the type of process, and the location and type of issued detected, the optimization engineers will develop a test plan that may include invasive and/or non-invasive tests.
The first course of action is non-invasive testing which involves collecting process data during normal production, with no changes introduced by the optimization engineers. This is a useful testing approach, but it may not be able to isolate the individual process interactions needed to identify a specific process issue. If non-invasive testing doesn’t provide the needed results, invasive testing is employed.
“One of the most common tests is a step test,” Tran says. “Optimization engineers vary the independent variable – their “prime suspect” – up and down to see how it affects the dependent variable, the condition or parameter related to the process problem.”
While the word “invasive” sounds like this test involves a major process disruption, the testing is coordinated with the customer to limit the magnitude of changes made so that the process remains within the current product specifications. That means little or no off-specification product will be created during the testing.
The results of the step test are often represented visually in a performance curve showing the interaction of the two variables. In some cases the performance curve takes the shape of a V. The point of the V indicates that there is one optimum condition or operating point for the independent variable. A bowl-shaped curve indicates that there is a broad range of conditions for the variable that will provide satisfactory production performance.
The performance curves provide continued benefit at end of the problem resolution process. The pre-improvement curve can be readily compared to the post-improvement curve to confirm the success of the effort, or to point out the need for continued problem correction.
“The success rate of optimization engineers after the first test is generally very high,” Murphy says, “in the 80% success range. However, there are always cases where the first test plan generates more questions than answers. It is sometimes, therefore, an iterative process, requiring a reconfiguration of the data gathering or a second attempt at correlating the data to the production issue.”
Check back next week for the “second way” and as always, we look forward to your comments.