A predictive maintenance program startedat the Cargill Vitamin E Plant in Eddyville,Iowa, has yielded hundreds of thousands ofdollars in documented savings since 2001through improved maintenance resulting indecreased process variability and greaterreliability. Estimated savings through costavoidance are even greater. For example,costly plant shutdowns have probably beenprevented by daily monitoring of 1500instruments and control valves, and thenscheduling maintenance based on criticali-ty of devices.

This is in line with a stated goal of find-ing cost savings through asset optimizationand predictive maintenance, which werenot previously practiced at the instrumentlevel in this plant. Prior to 2000, schedule-based preventive maintenance was sup-ported by reactive maintenance. We fol-lowed manufacturers’ recommendations incaring for their products, and we reacted toproblems as they arose. As long as thingsran smoothly, nothing was done to seek outpotential problems and prevent them frombecoming actual.

Predictive maintenance is the term givento the practice of identifying productionequipment needing maintenance attentionbefore its performance gets to the point thatproduct quality is reduced or an unplannedshutdown occurs. Information about the con-dition of equipment is the key to successful-ly predicting maintenance requirements inorder to take timely corrective action.

The information we rely upon principallycomes from smart, microprocessor-basedfield instruments that evaluate their owncondition and that of associated equip-ment in addition to reporting the basictemperature, pressure, and flow data need-ed to control production processes. Evenwith these advanced devices in place, tech-nicians formerly had to go into the plant,locate an instrument, and attach a hand-held communicator to extract the informa-tion. The evolution of communications pro-tocols, such as HART and FOUNDATIONFieldbus, and asset management softwaremade it possible for engineers and techni-cians to communicate with smart fieldinstruments from one central location.

Built in 1996, the Cargill Vitamin E Planthas nearly 1500 HART technology smartfield devices connected to its distributedcontrol system. In addition, digital valvecontrollers, or smart positioners, areinstalled on about 350 control valvesthroughout the facility. Massive amounts of

diagnostic data generated by these devicesare captured by the AMS Suite: IntelligentDevice Manager software installed nearlythree years ago. The information obtainedis integrated into a single database, organ-ized, processed, and presented on a PCmonitor, all without interfering with control

Solutions

Predictive Maintenance SavesMoney, Improves Plant Reliability

Cargill's Eddyville, Iowa facility.

system inputs and outputs. Using this tool,technicians are able to look deeply into theprocess in a way never before possible.From the safety and comfort of the mainte-nance shop, they’re able to perform all thefunctions generally associated with thehandheld communicator - instrument com-missioning, configuration, and trou-bleshooting of potential problems. StatusAlerts inform them if the performance ofspecific instruments begins to slip. In addi-tion, all maintenance activities are automat-ically documented.

Asset ManagementTo ensure that this software is fully utilized,we employ an asset manager, whose job it isto monitor the instrumentation, keep allconfigurations and calibrations current, andmaintain the database. Based on informa-tion received from the field instrumenta-tion, the asset manager determines mainte-nance needs and generates work orders asappropriate. He helped establish the pre-dictive maintenance system and trainsother maintenance personnel to use it.

He is also responsible for continuouslymonitoring the health and status of the 30Critical Control Points identified as essen-tial to the safety and quality of our product.Monitoring these critical points for baselineand calibration certification using NIST-traceable procedures allowed changing thequality control procedure from fully analyti-cal to a process instrumentation-based sys-tem. The reduction of in-process testing hassaved us enormous amounts of analyticaltesting time and dollars.

It is essential in any processing plant thatone individual be designated as asset man-ager. That person must be very knowledge-able about the process and the instrumen-

tation used for process control as well asthe intelligent device management soft-ware. This resource is too valuable to beignored, which could happen if a specificperson does not have this as a full timeresponsibility with authority to take whatev-er action is necessary to protect the integri-ty of the processing system.

Predictive MaintenanceEmerson’s Intelligent Device Manager soft-ware makes predictive maintenance possi-ble. Once the process control system isoptimized and all the loops are running aswell as they can possibly run, monitoring ofthe field instrumentation helps sustain thatlevel of efficiency. We now catch potentialproblems before they can impact quality orproductivity. A case in point is a recent trav-el deviation alert on a control valve thatcaused technicians to check out the valve,which seemed from the control room to beoperating properly. They found a plastic airsupply line that was too close to a steamline had melted and collapsed, restrictingthe passage of air to the valve actuator. As aresult, the valve was responding very slowlyto control signals and introducing a lot ofvariability into the process, but the opera-tors didn’t know anything was wrong.

In another instance, technicians checkingout the report of a “non-responding” instru-ment found that it had been disconnectedat the termination panel. This “fail-open”valve was allowing the loss of nitrogen,which went unnoticed by operators for sixmonths. While the process was not affected,fixing the problem prevented the furtherloss of nitrogen and reduced the loading ona scrubber by about 20 percent.

Many potential problems are simplyavoided because the asset manager system-atically checks every instrument on the net-work, looking for the warning signs thatindicate an impending failure. When hecatches a transmitter before it gets too farout of calibration or expires, he is helping tomaintain the reliability of the productionsystem. If early action by the asset manageravoids a plant shutdown, the savings couldbe substantial.

Since the predictive maintenance pro-gram began functioning, the reduction inwork orders has been noteworthy, withmaintenance following a more orderly planand schedule. Previously, a great deal oftime was spent reacting when somethingbroke. We’re still doing reactive mainte-nance, but now we’re reacting to predictionsbased on field-based information and notbreakdowns. Instead of smoke coming outof a motor housing, we react to an indica-tion that there will be smoke unless some-thing is done to prevent it.

Documentation BenefitsNIST-traceable calibration certificates mustbe provided as a part of the regulatory doc-umentation in the Vitamin E facility. Withour system, a calibration certificate on anyinstrument on the control network can beprinted out showing that instrument’s cali-bration history, accuracy, date last tested,condition “as found”, condition “as left”, etc.

To calibrate a transmitter, all the datarequired is downloaded from the IntelligentDevice Manager database to a documentingcalibrator, which is taken to where the actu-al calibration is done. At the end of the day,the results are uploaded to the PC andarchived in the database. Some instrumentsdo not have to be calibrated periodically aslong as we can verify that their calibrationparameters are within design specifications.The documentation within the instrumentdatabase serves as a reliable source for thatverification, making these historiesextremely valuable.

ConclusionIt is difficult to quantify the actual value ofpredictive maintenance in dollars. Savingsin some cases, such as preventing the lossof nitrogen, are easily computed. If variabil-ity on a steam loop is decreased, the energysaving can be calculated, but how can youput a value on decreasing loop variability?Similarly, it is impossible to assign a dollaramount to time saved by technicians - oftenbecause no data is available for compari-son. We are satisfied to say that enough hasbeen saved to pay for the software.

A word of caution; do not expect immedi-ate results! It is important for the individu-als working with the software to understandit and learn how to recognize potentialproblem areas. A tool is valuable only if it isused properly. If a predictive maintenanceprogram is implemented, an asset managerput in charge, and work practices changedas needed, savings should become appar-ent within 12 to 24 months.

Our work practices have certainlychanged. We have been able to shift ourfocus to prevent from correct, and the docu-mented savings are significant. We intend tokeep exploiting the predictive maintenanceenvironment and avoiding unexpectedstoppages. The more we can plan andschedule our work, the more efficient we willbe, and that is our ultimate goal.

— By Wade HowarthAutomation ManagerCargill Vitamin E PlantEddyville, Iowa

Wade Howarth, Automation Manager at the CargillVitamin E plant in Eddyville, Iowa, has helped hiscompany achieve "hundreds of thousands of dollars indocumented savings" over the last three years by apply-ing a predictive maintenance strategy based on assetmanagement software.

Reprinted with permission from Pharmaceutical Processing May 2004