Helena Kortelainen, Saku Pursio, VTT Automation Availability performance stands for plant efficiency ABSTRACT Availability performance is an efficiency measure, which can be calculated for any item from a piece of equipment to a whole production system. As input data, mean times between failures, mean times to repair and failure criticality are required for each item for identification of the parts, which contribute most of the system unavailability. Detailed availability performance analysis of a production system requires a model, which takes into account the logical structure of the system and the storage capacity between process stages. The model can also take into account the effect of production disturbances and planned stoppages if corresponding data is available. A comprehensive availability performance model is a practical tool when improvements are to be directed effectively, strategies tested, or alternative investement proposals compared. TIIVISTELMÄ Käyttövarmuus laitoksen tehokkuuden mittarina Sellu- ja paperiteollisuudessa tuotantolinjojen, koneiden ja laitteiden korkea käyttövarmuus on merkittävä tekijä pyrittäessä ylläpitämään ja parantamaan laitosten tuottavuutta. Käyttövarmuus Kilpailutekijänä teknologiaohjelman projektissa Tuotantolinjan käyttövarmuuden kokonaismalli kehitettiin käytettävyyden laskennan periaatteet ja käyttövarmuuden mallintamismenetelmä, jota sovellettiin paperin, kartongin ja sellun tuotantolinjoille. Projektiin osallistuvat VTT Automaation lisäksi Andritz-Ahlstrom (aik. Ahlstrom Machinery), Metso Paper (aik. Valmet), UPM Kymmene, Metsä-Serla Savon Sellu sekä ABB Industry. Käyttövarmuuteen liittyvä terminologia on vakiintumatonta ja se vaihtelee eri teollisuuden alueilla. IEC-standardin määritelmän mukaisesti käyttövarmuudella tarkoitetaan kohteen kykyä suorittaa vaadittua toimintoa, kun ulkoiset edellytykset toiminnon suorittamiselle ovat olemassa. Käyttövarma laite tai laitos toimii siis halutulla tavalla aina tarvittaessa. Yleisessä tapauksessa käytettävyys voidaan laskea vertaamalla toteutunutta toiminta-aikaa suunniteltuun. Käyttövarmuuden analysointia varten kohde tuotantolaitos tai sen osa on mallinnettava. Projektissa kehitetty käyttövarmuusmalli perustuu toiminnalliseen kuvaukseen, johon on liitetty tarvittavat luotettavuustekniset kytkennät. Käyttövarmuusmalli ja sen Excel-sovellus on osoittautunut tehokkaaksi työkaluksi tarkasteltaessa monimutkaisen järjestelmän käytettävyyttä. Sen avulla voidaan tunnistaa käytettävyyden kannalta ongelmalliset alueet, sekä testata ja vertailla vaihtoehtoisia parannustoimenpiteitä. Tulosten perusteella muodostetaan käsitys teknisesti ja taloudellisesti kannattavimmasta ratkaisusta. Käyttövarmuusmallin hyödyntäminen edellyttää vika-, häiriö- ja seisokkitietojen hankintaa. Tietoja voidaan joutua keräämään useista lähteistä ja täydentämään asiantuntija-arvioin. Mikäli kirjattuja tietoja on käytettävissä usean vuoden ajalta, havaitaan esim. vika- ja seisokkitaajuuden systemaattiset muutokset trendianalyysillä. Trendianalyysin tulokset voivat osoittaa järjestelmän vanhentumista, jolloin uudistustarpeet tulevat selkeästi esille. Trendi voi osoittaa myös järjestelmän parantumista, esimerkiksi kulutusosien kestoiän pidentyessä. Käyttövarmuuteen liittyvä informaatio on tärkeää niin laitevalmistajille kuin laitoksen henkilökunnallekin joka on vastuussa laitoksen käytöstä ja kunnossapidon kehittämisestä. Käyttövarmuuden tunnuslukujen laskeminen yhteisesti sovittujen periaatteiden mukaisesti mahdollistaa benchmarking-tarkastelujen tekemisen erilaisten ratkaisujen ja laitosten välillä. Introduction Availability performance has become one of the most important factors in the competition between companies in the same market. High availability performance for process plants, machines and equipment means undisturbed and safe production, minimised maintenance costs and minimised off-grade production. All these facts improve the efficiency and the flexibility of the production. Both machinery suppliers and users need exactly defined measures for availability performance. Operating and maintenance personnel use availability performance data to support decision making at the plant, including both day-to-day work and investment planning. The objective is higher availability performance and the information obtained helps corrective measures to be directed appropriately. Machine and equipment manufacturers need availability performance data in the product design and specification phase, and also because the prospective users have begun to demand convincing reliability specifications and guarantees. This paper presents availability performance modelling and data analysis methods developed in the Finnish Nation- 292 Paperi ja Puu Paper and Timber Vol.83/No. 4/2001
al Research Programme Competitive Reliability /1/. The project partners included VTT Automation, the pulp & paper industry companies UPM-Kymmene and Metsä- Serla, and machinery manufacturers Andritz-Ahlstrom (formerly Ahlstrom Machinery), Valmet and ABB Industry. What is availability performance? Reliability, availability and lifetime planning have already reached advanced levels in the nuclear and aerospace industry. In the process industry, availability is often a more important criterion than reliability. Even a sudden failure may be acceptable if the repair and restarting times are short. According to the international standard /2/, availability performance is the probability that a system is in a state to perform a required function under given conditions at a given instant of time or time period, assuming that the required external resources are provided. A shutdown may be caused by a failure, by a maintenance or production action, by lack of external resources or by other reasons. These causes of downtime have to be taken into account together with the standard definition when the mathematical expressions for availability performance are defined /3/. As the availability performance figures should provide support for decision making in different situations, more than one definition seems to be necessary. Timebased overall efficiency (Eq.1) is a typical efficiency measure applied in the pulp and paper industry. It also describes the overall availability performance of the production system. Downtime (maintenance) refers to the downtime caused by corrective and preventive maintenance actions. Eq. 2 does not differentiate between planned and unplanned maintenance actions, as both reduce the usable production time of the machine. For pulp and paper manufacturers, the high availability performance of a technical system and low overall efficiency indicate that the machinery is doing well and that problems arise from the process or other process-related factors. Typical process-related factors causing downtime are machine washing and felt changes. Low technical availability performance turns the focus of attention to maintenance and to the performance of hardware items. Machinery manufacturers need availability performance data in product design and also because the prospective users have begun to demand convincing reliability specifications and guarantees. Calculation of the availability performance figures according to mutually agreed definitions allows benchmarking and comparisons between different designs and plants. Availability performance models Modelling principles Availability performance can be calculated for a hardware item, a function or a system. In order to be able to derive the system availability from the item availability figures, an availability performance model needs to be constructed. Once the model exists, the availability can be calculated analytically or by simulation. The ideas of functional modelling /5,6/ were utilised when developing availability performance models for pulp and paper production. Functional modelling offers a hierarchical top-down approach, which starts from the top function of the system. The top function is usually the reason the system was originally built, e.g. Produce paper. The essential part of the model is the logical structure, which not only defines the connections between sub-functions, but also how the availability of a hardware item influences the system availability performance /7,8/. The availability performance model developed here is schematically presented in Fig. 1. The data required for an availability performance calculation consists of the hardware item failure rates and repair times, preventive maintenance actions and the corresponding information on other shutdowns. Maintenance event history is usually recorded in maintenance management systems, but availability-related information is also collected in other systems such as automation and production management systems. Unfortunately, in many cases the recorded data is not detailed enough or is not in a suitable form /9/ and has to be supplemented by engineer judgements /10/. Productiontime Overall efficiency = Maximumusable productiontime (Eq.1) Production time and maximum usable production time are specified in a widely read publication by the German pulp and paper engineers society /4/. Repairs or maintenance actions cause only some of the production stoppages. The function of the machinery is measured by the availability performance of a technical system (Eq.2). Max.usableprod. time downtime(maintenance) Techn.availablity perf. = Maximumusable productiontime (Eq.2) Fig. 1. A schematic presentation of the hierarchical availability performance model /7/. Paperi ja Puu Paper and Timber Vol.83/No. 4/2001 293
The influence of storage capacity In the pulp and paper industry, some storage capacity (e.g. intermediate tanks) is usually placed between successive production stages. This intermediate storage capacity ensures a steady material flow through the whole system, thus reducing the probability of total production stoppages. However, storage capacity is not usually taken into account when modelling system reliability. In the availability model developed here, the tanks are considered to be functional units and the inability to perform the required function is regarded as a failure /7/. In terms of availability performance, the required function of a tank is to provide the next process stage with sufficient amounts of material. The preceding process stage has the same function and therefore a process stage and a tank can be regarded as one functional unit as shown in Fig. 2. In this example, process stage 2 receives material from the preceding stage 1 or from the tank. Production stops only if stage 1 is down for some reason, and there is no material left in the tank. The intermediate storage capacity may have a significant impact on the system level availability performance, as shown in the example in Table 2. Omission of the storages from a reliability model leads to underestimation of the system availability. However, intermediate tanks have diverse effects on both production efficiency and product quality. In some cases, better results could be achieved by improving the availability performance of the machinery instead of incorporating a large tank into the process. The effects of the two possible approaches can be tested using the availability performance model. The model requires data on tank volume, liquid level and level variations. At an existing plant this information can usually be derived from the process automation system. The model interface The people responsible for operation and maintenance management and for machinery and process development, as potential users, profit from availability performance models. Accordingly, a commercial spreadsheet program was chosen and the availability performance modelling was performed using MSExcel. The model interface is easy to use and does not require installation of new software or comprehensive training. The interface of the MSExcel application is shown in Fig. 3. The availability performance models and the corresponding model interfaces are based on a functional description of the Fig. 2. Process stage and tank as one functional unit. Fig. 3. Availability performance model - interface of the MSExcel application for a pulp mill. The figures are based on made-up data. pulp and paper processes. As the functions and basic process steps are practically the same at most pulp and paper mills, the models developed can be applied in the pulp and paper industry in general. Some modifications may be required as the hardware varies from machine to machine. However, the modifications are fairly easy to make. The modelling principles and modelling tool were also successfully applied when creating an availability performance model for the paper machine s electricity system /11/. The input data consists of failure rates and repair times for hardware items and the corresponding data on other shutdowns. The model also requires information concerning the failure consequences (no effect / stops the item / stops the system). Availability-related trends At some mills, shutdown and failure data has been systematically collected for years not necessarily in the maintenance management system but by other means. Variations in failure or shutdown rate can easily be studied by plotting the cumulative number of events as a function of cumulative time. The main limitation in using the simple statistical method is the amount of data. The minimum number of events for reliable results is ten, and the more data the more reliable are the results. The variation may be random or periodic, or it may show a clear increasing or decreasing trend /12/. An increasing trend indicates a deterioration of the system and a decreasing trend system improvement. Fig. 4. The failure rate of an electric system shows an increasing trend and the shutdown frequency due to felt changes shows a decreasing trend. 294 Paperi ja Puu Paper and Timber Vol.83/No. 4/2001
Random variation suggests that the system is neither improving nor deteriorating. Two examples in Fig. 4 present clear trends: an increasing failure rate of the electric system and a decreasing shutdown frequency of the felt changes. The electric system in Fig. 4 shows typical ageing: the failure rate starts to increase after a long constant failure rate period. The increasing failure rate also adversely affects the plant availability performance (see Table 1), and a replacement of the old electric system with a modern one therefore seems to be a justifiable investment. The other example a decrease in the number of shutdowns caused by felt changes may be a consequence of several measures such as increased washing rate, improved felt durability or changed process parameters. In such case, a detailed analysis of the possible causes of the change would be needed. The trend analysis itself does not produce any predictions, but points out the direction of change. The trend analysis is a simple and concrete way to control the performance of a subsystem or an individual piece of equipment, as the deviations from normal behaviour are readily seen. In order to get full advantage from the analysis, the trends should be viewed together with other relevant information, e.g. changes in the process characteristics or machinery. The failure rates derived from the trend analysis can be further used in the availability performance models. Availability performance in decision making The availability performance model allows the user to look for the bottlenecks in production and hardware items causing most of the downtime. The procedure is illustrated in Fig. 5. The model offers assistance when estimating the effect of a planned change. A typical change is a replacement or modernisation investment: outdated machinery or part of it is replaced with the latest engineering. The interesting issues are: how much the system will be improved by the modification and whether the investment will be profitable. Two examples in Table 1 show how much a replacement or modernisation investment increases the system availability performance. Real plant data was used in calculating the examples. Trend analysis was used to get the shutdown rates, and the availability performance model was used in calculating the system availability. Proposed investments increase the plant availability performance, while the failure Fig. 5. Which part of the system contributes most to technical availability performance? Calculations are based on hypothetical data. Table 1. Replacement and modernisation improves availability performance case examples. Board machine: an aged electrical system is replaced with a new one Paper machine: tail threading ropes are replaced with an air shower tail threading system Table 2. Influence of storage capacity on plant level availability. rate and downtime are reduced. The economical impact is significant, as even the 0.3% increase corresponds to one more day of running time per year. The availability performance model also allows the corresponding comparison on cost terms and the results can be further analysed, e.g. using life cycle profit and cost analysis /13/ for profitability analysis. The system efficiency can be increased by increasing the intermediate storage capacity or by improving the availability performance of the machinery. The model offers a tool to estimate the benefits and System availability increase +0.3 % units +0.7 % units Technical availability performance of a pulp mill, % A pulp mill with no intermediate storage capacity 76 A pulp mill with existing intermediate storage capacity 93 A pulp mill with unlimited intermediate storage capacity 95 consequences of the two alternative approaches. Also, different production parameters can be tested, e.g. an adjustment of storage capacity or tank surface level. Some examples are shown in Table 2. At a pulp mill, tanks ensure a steady production flow. However, the mill s availability performance does not reach 100% even with unlimited storage capacity, as the last production step the drying machine reduces the system availability. The figures in Table 2 clearly show the significance of the availability aspects in process design and in choosing the process machinery. Paperi ja Puu Paper and Timber Vol.83/No. 4/2001 295
Conclusions The availability performance model developed here has proven to be a useful tool for analysing the availability performance of a complex system. Bottlenecks in production, and the hardware items causing most of the production downtime can be identified and this helps the plant personnel to direct the limited resources more effectively. Alternative improvement strategies can be tested and compared easily. The user can choose the solution that is technically and economically most profitable. Testing is not restricted to existing systems but is also very useful in the process planning phase, when it is easiest and cheapest to make modifications. Changes in failure rates are detected with trend analysis. The results may indicate a general deterioration of the system and be a signal of coming need for costly renewals. Trends may also show system improvement, e.g. due to the increased lifetime of consumables. Trend analysis is basically a simple method and could be incorporated into a maintenance data collection system. Availability-related information is important both for the plant personnel responsible for operation and maintenance development and for the machinery manufactures. Calculation of the availability performance figures according to mutually agreed definitions allows benchmarking and comparisons between different designs and plants. References 1. Holmberg K., 2000. Finnish research focuses on competitive reliability. Kunnossapito 2/2000, pp 14 31. 2. IEC 50(191), 1996. Electrotechnical vocabulary. Dependability and quality of service. 143p. 3. Kortelainen H., 1999. Käyttövarmuuden mittarit. Tampere. VTT, RISB005. 22. 4. Merkblatt II/2/81 1981. Produktionskennzahlen für die Papiererzeug. Verein der Zellstoff- und Papier-Chemiker und Ingenieure. Darmstadt. 8p. 5. Ross D., 1985. Applications and Extensions of SADT. IEEE. April 1985, pp 25 34. 6. http//www.idef.com 28.10.2000. 7. Pursio S., 1999. Tuotantolinjan käyttövarmuusmalli. Diplomityö. Tampere. 85p. 8. Kortelainen H., Salmikuukka J., Pursio S., 2000. Reliability simulation model for systems with multiple intermediate storages. Annual Reliability and Maintainability Symposium Proceedings. Los Angeles CA, Jan. 24 27, 2000. pp. 65 70. 9. Konola J., Mäki K.M., 1999. Käyttökokemustiedon keruun tarpeet ja mahdollisuudet. Tampere. VTT, RISB001. 12p. 10. Kortelainen H., Ristimäki P., Oinonen K., 1998. Paperikoneen toiminnallinen kuvaus ja käyttövarmuusmalli (Functional description and availability performance model of a paper machine). VTT Symposium 188. Espoo. 19.11.1998. Holmberg K.(ed.). VTT. Espoo, pp. 87 93. 11. Huuhko H., 2000. Paperikoneen sähkökäytön käyttövarmuus. Diplomityö. Espoo. 102p. 12. Bohoris G., 1996. Trend testing in reliability engineering. International Journal of Quality & Reliability Management. Vol.13, No. 2, pp. 45 54. 13. Taipale, Ville, Peltonen, Minna, Rouhiainen, Veikko 1998. Elinjaksokustannusten ja -tuottojen tarkastelu investointipäätösten tukena. Tarkastelukohteena Safematic Oy:n voitelu- ja tiivistämisjärjestelmät. Kunnossapito 8/98 s. 40 42. 296 Paperi ja Puu Paper and Timber Vol.83/No. 4/2001