White Paper Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid Executive Summary In tday s prcess industries, the plant flr is awash in infrmatin. Harnessing that data - which means rganizing and making it usable - is what yields meaningful perfrmance gains. Cmpanies ften use dzens f applicatins t manage cmplex prductin peratins, mnitr prcesses and make perating decisins. These systems are usually either cmpletely islated r cnnected with cmplex, custm-designed interfaces that make it difficult t use the data effectively and maintain its integrity. The simple measurement f plant perfrmance metrics can be a daunting task. Perfrmance data is frequently lcked up in disparate systems and must be nrmalized befre analysis. Finally, mst cmpanies cntinue t rely n basic spreadsheet applicatins fr metric tracking, limiting the ability t analyze large sets f data. This whitepaper describes hw t design an effective perfrmance management system and avid sme f the cmmn pitfalls..
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 2 Table f Cntents Backgrund...3 What is Perfrmance Management?...3 Understanding the Rle f KPIs. 4 Critical Infrmatin Requirements...7 Seven Pitfall t Avid in Perfrmance Management...8 Benefits t End Users....9 Cnclusin....10
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 3 Backgrund Manufacturing cmpanies emply a variety f infrmatin systems t allw emplyees t d their jbs. But hw easy is it fr different rles t access critical prductin data? Each wrker is fcused n the key perfrmance indicatrs (KPIs) fr his r her individual functin. These perfrmance targets and measurements are imprtant fr aligning emplyee decisins and actins with verall business bjectives. They can range frm safety, peratins and reliability, t ecnmics, energy and lss, plant expansin and persnnel. Fr example, peratrs mnitr thrughput and cycle time t keep prductin prcesses running cntinuusly, while the maintenance team mnitrs asset perfrmance and tries t predict and prevent dwntime. Plant managers must keep their attentin fcused n prfitability and utilizatin. Traditinal rle-based infrmatin, held in islated sils, hasn t been accessible t mnitr the state f verall plant peratins. In cases when data is shared amng different departments, it is ften distributed n spreadsheets cntaining manually keyed infrmatin. Nt nly des this pull peple away frm the tasks at hand, it als intrduces human errr int the prcess. Many industrial rganizatins are discvering the value f harvesting data frm the plant flr t respnd t KPIs that influence imprtant decisins at the highest levels. Data left drmant and islated in disparate cntrllers, human-machine interfaces (HMIs) and ther enterprise systems can prvide vital infrmatin abut verall prcess efficiency, system uptime, energy usage, cst f materials, envirnmental cmpliance and ther KPIs. Withut questin, metrics matter when it cmes t ptimizing peratinal and business perfrmance. What is Perfrmance Management? Yu can't manage what yu dn't measure. It is an ld management adage that is accurate even tday. Unless yu measure smething yu dn't knw if it is getting better r wrse. Yu cannt manage fr imprvement if yu dn't measure t see what is getting better and what is nt. Fr this reasn, manufacturers cllect data (measurements), determine hw thse will be expressed as a standard (metric), and cmpare the measurement t the benchmark t evaluate prgress. Fr manufacturing peratins, an apprpriate and cnsistent set f easily-understd perfrmance metrics needs t be within quick access. This requires cmprehensive metrics framewrks and autmated management capabilities. Almst every rganizatin has put in place metrics prgrams f sme kind. Taking the next step t a cmprehensive system fr managing enterprise metrics has been daunting fr a number f reasns: Metrics are a mving target. Visibility int metrics is uneven. There are n cnsistent framewrks f essential metrics. Analysis f data interrelatinships s vital t decisin making is cmplex.
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 4 Creating KPIs and aligning them with rganizatinal bjectives is key in establishing an effective mnitring system. A screcard is a ppular term fr describing such a mnitring system, and mapping it t individual respnsibilities. Dashbards are particularly effective in delivering a screcard t a wide audience. It is cmmn t expect that different jb rles and functins within an rganizatin will require different sets f metrics and dashbards. Creatin f a diverse and rich cllectin f KPIs makes dashbards an essential driver fr change and psitive transfrmatin. Understanding the Rle f KPIs KPIs are metrics that are used t assess the current state f a business, and t measure prgress twards gals. Fr a manufacturing facility such as a refinery, KPIs are useful t measure perfrmance in each benefit area. There are many pssible KPIs, and each plant will need t cme up with KPIs that are apprpriate t the way it runs. This paper refers t primary KPIs as metrics that directly measure perfrmance in a benefit area. Fr instance, unit feed rate is typically a primary KPI because higher feed rate usually means mre revenue and prfit. Secndary KPIs are measurements that supprt r explain primary KPIs, and thus have an indirect influence n benefits. Tertiary KPIs are measurements that supprt r explain secndary KPIs, but d nt have a direct r bvius relatinship n benefits. These are lse definitins, and a metric that is a primary KPI at ne plant can easily be a secndary KPI at anther plant. It is als true that sme metrics simply d nt apply at sme facilities. Fr instance, a plant that has plenty f spare capacity may decide that metrics based arund thrughput d nt mean much. Each plant must cme up with its wn set f KPIs that are useful fr it. Ptential benefits can be estimated in several ways: The benchmark methd cmpares current perfrmance t sme external benchmark r standard. Benefits are estimated frm the difference between actual and target perfrmance. The similar plant methd cmpares current perfrmance with similar plants. Benefits are estimated frm differences in metrics at the tw plants. The variability methd analyzes the variability in histrical data. Benefits are estimated frm reducing the variability, thus allwing perating mre cnsistently at a mre desirable target. Experience. The fllwing table presents sme benefit areas and metrics fr a typical refinery. Metrics fr ther industries will differ but the general type and rganizatin f metrics will be similar. These metrics are generally primary KPIs, which have a direct measurable link t business value and therefre imprving the KPIs will nrmally result in measurable benefits. The last clumn shws the methd recmmended fr estimating ptential future benefits frm histrical data. Area Prcess Unit Metric Estimate Benefits Using Yield Management & Capacity Utilizatin All Effectiveness = actual/target feed rate Variability, valued at grss margin All Variability All Unit material balance Benchmark fr mass balances, Variability fr vlume balances CDU Actual/target atmspheric uplift (AGO and lighter) Variability, valued at yield shift
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 5 Area Prcess Unit Metric Estimate Benefits Using Prduct Cst and Quality Inventry, Lsses, and Wrking Capital Lgistics & Transprtatin Fixed & Variable Operating Csts Energy Efficiency Reliability & Plant Perfrmance Safety & Regulatry Cmpliance Staff, System Prductivity Refrmer Actual/target C5+ yield Variability, valued at gasline natural gas spread FCC Cnversin = (feed cycle stck)/feed Variability FCC Selectivity = naphtha yield/feed Variability Hydrcracker Cnversin = (gas thrugh ker)/feed Variability Hydrcracker Reactr WABT Variability Cker Gas yield/feed rate Variability Cker Actual/target drum utage Variability Offsites Octane giveaway in gasline pl Variability Offsites Gravity, cetane, sulfur, and pur pint giveaway Variability Offsites Effectiveness = measured/target prductin Refinery Unaccunted lss as % f thrughput Variability, valued at cst f crude il Refinery Flaring as % f thrughput Benchmark, valued at fuel r alternate use Refinery Days f crude il supply Variability. Difference between actual and lwest sustained inventry, valued at cst f crude il Refinery Refinery Slmn calculatin f ttal maintenance spending Slmn calculatin f catalyst, chemicals & industrial cnsumables Benchmark - difference between actual and Slmn target Benchmark - difference between actual and Slmn target Refinery Slmn target fr energy intensity Benchmark - difference between actual and Slmn target, valued as % f ttal energy budget Prcess units Refinery, each prcess unit Each cnsle Refinery Refinery, each prcess unit Refinery Actual energy use / target usage adjusted fr actual prcess cnditins Service factr, unplanned slwdwn r dwntime, r dwntime using OEE definitins Alarm metrics cmpared t ASM r ther benchmarks Safety statistics, such as OSHA s Recrdable Incident Rate (IR) Emissins, such as quantity flared r cnfrmance t EPA standard 40 CFR (Prductivity metrics tend t be specific t individual plants) Variability, valued at marginal fuel Benchmark use actual dwntime valued at refinery grss margin Benchmark Benchmark Benchmark Benchmark
Tertiary KPIs Secndary KPIs Primary KPIs Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 6 The primary KPIs shwn abve will be supprted by secndary and tertiary KPIs. Primary KPIs relate directly t business results, while secndary and tertiary KPIs are easier t understand hw t imprve. The next table shws a typical set f KPIs that relate t the yield management and capacity utilizatin benefit area fr a typical prcess unit, in this case an FCCU. Type Metric Measures t Mnitr Unit feed rate Unit mass balance, unit vlume balance Gas yield Naphtha yield (debutanizer bttms) Cycle stck yield (LCO and heavier; prtin nt cnverted t naphtha) Cnversin = (feed-cycle stck)/feed Selectivity = naphtha yield/feed Efficiency = selectivity * cnversin Advanced cntrl utilizatin Active cnstraints Catalyst circulatin rate Catalyst/il rati Prpylene yield (C3 splitter verhead) Prpane yield (C3 splitter bttms) Flash zne temperature Riser utlet temperature Regenerated Catalyst Slide/Plug Valve Differential Pressure Regeneratr cyclne temperature Regeneratr O2 Air blwer discharge pressure Wet gas cmpressr inlet pressure Main fractinatr differential pressure Naphtha (debutanizer bttms) 90% pint Main fractinatr LCO 90% Analyzer availability Analyzer repeatability Absrber clumn differential pressure Debutanizer clumn differential pressure Debutanizer verhead C5+ Naphtha (debutanizer bttms) RVP r C4- Measured, actual value Actual/target (effectiveness) Actual/capacity (utilizatin) % gain/lss % unit feed rate % unit feed rate % unit feed rate % time APC is active Number active APC cnstraints % time selected CVs are active cnstraints % unit feed rate % unit feed rate Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC % time unit analyzers are nline % f lab sample results within nrmal precisin f lab and analyzer results Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 7 Type Metric Measures t Mnitr Naphtha (debutanizer bttms) ctane Deprpanizer Bttms C3- Deprpanizer clumn differential pressure Deprpanizer verhead C4+ cntent C3 Splitter verhead C3 cntent C3 Splitter bttms C3 lefins cntent C3 Splitter clumn differential pressure Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC Inferred prperty calculatin, as emplyed fr APC Critical Infrmatin Requirements Manufacturers knw that t be cmpetitive in tday s marketplace, they need access t the right infrmatin, at the right place, at the right time. This requires the tls t make better decisins by eliminating guesswrk and managing what is measured. It is nt acceptable t find ut tmrrw that yu shuld have changed sme things yesterday t imprve prfit margins. Plant managers at industrial facilities face a hst f perating challenges: Hw d I better manage my assets? Hw d I ensure safe and stable peratins? Hw will demand influence my prductin? Hw can I cmply with industry and gvernment regulatins? Hw can I enable my wrkfrce t achieve imprved perfrmance? Prcess industry peratins arund the wrld struggle under the sheer vlume f data they generate. There is simply t much infrmatin t prcess, understand and act n quickly, leading t pr decisin-making. With the pwer t access the right data when and where it s needed mst and cllabrate acrss business units, users can take infrmed actins t achieve peratinal excellence. At mdern industrial facilities, it s nt uncmmn t find a hdgepdge f infrmatin that des nt allw a cmplete picture f plant perfrmance. Separatin and segregatin f departments is als a hindrance t cmplete infrmatin and display, bth within a site and ver multiple sites. Althugh technlgical barriers have been remved between autmatin, peratins and maintenance, business practices enabling cmpanies t truly benefit frm this newfund freedm have been slw t change. Breaching the walls can allw cmmn access t flds f data, but typically desn t imprve cmmunicatin. It is mre imprtant than ever t find a slutin fr managing plant metrics. But part f the task f implementing these metrics is determining hw they are defined and calculated. This necessitates the invlvement f peple frm a number f departments, wh may have had little cntact previusly, t reach cnsensus n the metrics requirements. At the plant level, rganizatins need data cllectin and reprting tls t assemble data n equipment status, prcess perfrmance, resurce cnsumptin and ther critical KPIs. Once captured, this data must be prperly cntextualized based n perating parameters that render the infrmatin either significant r insignificant. Plant managers, maintenance departments and peratins staff then need the ability t view perfrmance data in a meaningful way t determine which KPIs are being achieved within their span f cntrl.
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 8 Seven Pitfalls t avid in Perfrmance Management The effectiveness f the perfrmance management prgram based n the KPIs relies n varius factrs. In rder t have an effective perfrmance management prgram using KPIs, sufficient care must be taken thrughut the prgram t lk ut and eliminate the fllwing seven pitfalls. 1. KPIs are nt really KPIs: Many times KPI s are cnfused with measurements r metrics. While all KPI s are metrics, all metrics are nt KPI s. A gd KPI has fllwing characteristics Well defined The KPI must be defined withut ambiguity and in clear terms. Cnsistent The KPI need t be applied cnsistently acrss functins/departments. Measurable r cmputed accurately The KPI must be measurable, and any cmputatin dcumented clearly. Respnsive t change The KPI must be able t indicate result f any actins being taken. Timely - The KPI must be fr apprpriate time windw like daily, weekly r mnthly. Owned The KPI must be assigned a clear wnership t a specific rle/functin. 2. Everyne wants everything: There is a general tendency in rganizatins t prvide available data t mst users as nt prviding the infrmatin may be cnstrued as lack f infrmatin availability. But it verlads the user and the mst relevant infrmatin is buried in the pile f infrmatin. It is imprtant the mst relevant and actinable infrmatin fr that particular user is presented in the mst meaningful frm first. A gd strategy fr this is t rganize infrmatin in frm f dashbards and drill-dwns. The main dashbard shuld cntain key infrmatin and highlight exceptins easily. The drill dwns shuld cntain any detailed infrmatin. There is significant benefit in wrking with users t make the transitin and t ratinalize reprting and infrmatin delivery 3. Nt enugh leading indicatrs : Particular attentin shuld be paid t leading indicatrs. Fcus n KPIs that are actinable nt thse reprting histry. Avid t much emphasis n Lagging indicatrs. A Lagging indicatr tends t be a measured utput. A Leading indicatr is a measure that predicts hw anther measure, ptentially a lagging indicatr, might behave in the future. Leading indicatrs let yu react t influence a lagging indicatr. Safety incidents may be influenced by Number f High risk jbs being perfrmed Bypassed safety equipment Fllw-up safety audits nt cmpleted Financial cst perfrmance may be impacted by Number f cst saving pprtunities identified but nt yet implemented 4. Lack f data quality: Lack f cnfidence in the numbers can seriusly undermine a KPI prgram. Data Surces shuld use standard maintainable integratin methds OPC, Web Services. Access methds need t perfrm and prvide timely, reliable infrmatin. Identificatin f stale data is imprtant. Need t accmmdate recalculatins fr crrectins r late arriving data. Master data needs t be managed t ensure systems are in sync 5. Incnsistencies acrss rganizatin: Lack f cnsistency acrss the rganizatin shuld be avided. Step ne is t identify that yu have incnsistencies. Requires an audit f the KPIs, calculatins and data cllectin methds Beware the manual entry spreadsheet T ensure cnsistency drive yur dashbards as mdel-driven templates Drive KPI calculatins the same way Implement Management f Change prcess Where Excel is still required apply cntrls 6. One and dne: All the effrt spent n creating a KPI prgram can be wasted if attentin is nt give t Change management prcess Hw t effectively intrduce changes in business prcesses
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 9 Sustainment Ensure prcesses are in place t maintain the systems and data Supprt fr users Prvide supprt fr user in using the system Cntinuus Imprvement Ensure the system is flexible enugh t evlve. Empwer Users Empwer users t make their wn changes t tune the system t meet their specific needs 7. Dn t d anything: Nt ding anything is the biggest pitfall. Effective perfrmance management while needs f attentin and care t be successful, the benefits are immense and it allws rganizatins t take a step change in their current perfrmance. Hence, it is critical t institutinalize perfrmance management and btain business benefits. Pitfall Recmmendatin 1 KPIs are nt really KPIs KPIs need t be aligned with business gals, be measureable and actinable 2 Everyne wants everything Deliver cncise exceptin based dashbards with drilldwns t details as required 3 Nt enugh leading indicatrs Fcus n leading indicatrs that can allw actin befre prblems ccur 4 Lack f data quality Ensure data quality with standard interfacing and data management prcesses 5 Incnsistencies acrss rganizatin Enfrce cnsistency acrss the rganizatin with standards and templates 6 One and dne Empwer users and set up management f change prcesses t facilitate cntinuus imprvement 7 Dn t d anything Define the KPIs t drive business benefits nw and fr the future Benefits t End Users A well-defined perfrmance management prgram can help a manufacturer imprve its prductin efficiency, reduce perating csts, increase thrughput, and imprve cmpliance with gvernment and industry standards, as well as ensure gd staff mrale. Emplyees want t d their best and it is imprtant t ensure they have the right tls and spnsrship. They als need t knw hw they are perfrming and where they can imprve. Autmatin f metrics management can help users capture metrics/kpi data, perfrm analysis against the minimum threshlds they have set, autmatically issue alerts when warranted and deliver scheduled reprts t subscribers. In essence, mst f the prcess can be autmated s the administrative burden in managing a KPI/metrics prgram is kept t a minimum.
Perfrmance Management in Prcess Plants: Seven Pitfalls t Avid 10 Cnclusin A grwing number f industrial rganizatins are implementing cllabrative strategies intended t crrelate perfrmance metrics with real-time KPI infrmatin. Implementing an effective KPI management system will require aviding certain pitfalls and the same have been discussed in this whitepaper. Designing, implementing and maintaining an effective perfrmance management system results in bttm line benefits t an rganizatin. Fr Mre Infrmatin Learn mre abut hw Hneywell s Data, Analytics and Cllabratin can bring real-time digital intelligence t yur prcess plants, visit ur website www.hneywellprcess.cm/sftware r cntact yur Hneywell accunt manager. Hneywell Prcess Slutins Hneywell 1250 West Sam Hustn Parkway Suth Hustn, TX 77042 Hneywell Huse, Arlingtn Business Park Bracknell, Berkshire, England RG12 1EB UK Shanghai City Centre, 100 zunyi Rad Shanghai, China 200051 www.hneywellprcess.cm WP-15-10-ENG May 2015 2015 Hneywell Internatinal Inc.