Forest Energy Observer 27 Forecasting Algorithm for Natural Drying of Energy Wood in Forest Storages Lauri Sikanen, Dominik Röser, Perttu Anttila, & Robert Prinz Increasing procurement volumes of forest biomass for energy require improved information systems to control and manage transportation. Moisture content of the biomass is the most important quality parameter to follow. Transportation of water in solid biofuels increases both costs and CO2 emissions. Procurement operations need to be directed to storages which have low enough moisture content. Storing the fuel too long time increases capital costs of procurement as well as heating value losses. This complex decision making situation requires for possibility to forecast moisture content changes of storages as a part of the procurement control of energy wood. This study introduces one approach to create forecasting algorithm based on daily moisture change and drying periods. The algorithm is simple and easy to program and can be a part of enterprise resource planning (ERP) applications. Keywords: Fuel chips, biofuels, forest fuels, bioenergy, moisture, forecast Contact: Prof., Dr. Lauri Sikanen Mekrijärven tutkimusasema, Yliopistontie 4, 82900 Ilomantsi, Finland +358(0)2944 53683 e mail: lauri.sikanen@uef.fi
1 Introduction 1.1 Background Moisture content of solid biofuels from forest can vary from fresh wood 50 60%-MC down to 20%-MC without artificial drying (Hillebrant & Nurmi, Röser & Sikanen). Moisture is changing according to climate around wood storages and piles. Climatic factors affecting timber drying are commonly recognised to be evaporation and precipitation, which are a result of temperature, wind, relative humidity, radiation and rainfall. Annual changes of drying conditions are usually systematic, but there are a lot of differences between single years. Moisture content of the energy wood pile is the most important decision criteria for further procurement actions, chipping and transportation. Transportation managers need to make working programmes based on fuel demand and availability of fuel storages. In order to be able to include optimal storages into chipping and transportation programme, managers need to have information about the moisture content of piles. Measuring or sensoring of piles is usually too complex or expensive. Due to that, a forecasting algorithm in transportation management application is easy and cost effective alternative. 1.2 Aim of the Study In this study, simple and robust way to create forecasting algorithm for natural drying of energy wood will be demonstrated. 2 Material and Methods The most important data are all possible research reports dealing with natural drying of energy wood in certain climatic area. The idea is to get enough data so that the trends of the whole year can be estimated. In this case we are concentrating to central Finland. In Figure 1 is presented drying curves of energy wood by Nurmi & Hillebrand (2007). Effective drying during spring time is recognisable, as well as re-moisturing in the autumn. Winters in Finland are problematic, snow usually does not increase the moisture content of the wood, but if snow and ice are not properly removed during the chipping operation, they can increase the moisture content of the load. 2
Figure 1. Moisture changes of individual whole trees (WHT) and multi tree handled (MTH) bunches in in-woods storing in Töysä and Ruokolahti. (Nurmi & Hillebrand 2007). In Figure 2 corresponding figures are presented from the study of Röser et al. 2011 Figure 2. Moisture content changes of Finnish pine and alder in drying trial by Röser et al. 2011. Months on the X-axis. Gray line, partially debarked. 3
In Figure 3 annual fluctuation of moisture can be seen and also the effect of covering of piles is clearly recognisable. Figure 3. Moisture of April-felled and delimbed energy wood stems in different seasons of the year Hillebrand et al. 2005). When an adequate amount of drying data is obtained, the modeling of year-around drying process needs to be done. One approach can be statistical modeling, however, heuristic fitting can be the most appropriate approach in some cases. In heuristic fitting one must remember that algorithms need to be programmed into ICT applications and algorithms need to be flexible to be varied and adjusted for new climatic conditions. The use of algorithms also produce new information and increase the need of improved versions. 4
3 Results In this study, the year was divided in seven drying periods according to the speed of drying adapted from earlier studies. The factors describing the drying period is the length in days and the moisture change during the period. As a result, following description was produced (Table 1). Table 1. Drying period and moisture change estimates during one year in typical Central-Finnish forest storage of whole tree energy wood. Drying periods Start Stop Number of days Moisture change during the period (%-units) Daily moisture change (%-units) 2010-01-01 2010-02-28 58 0 0.000 2010-03-01 2010-04-01 31 3 0.097 2010-04-02 2010-06-15 74 12 0.162 2010-06-16 2010-07-30 44 5 0.114 2010-07-31 2010-08-31 31 1.5 0.048 2010-09-01 2010-11-30 90-12 -0.133 2010-12-01 2010-12-31 30 0 0.000 The most important factor is the daily moisture change (DMC). It tells the amount and direction to which moisture changes every day during the storage period. DMC is calculated by following formula: DMC = MC / ND Where: MC = Moisture change during the period, %-units. ND = Number of days in period DMC enables calculation anytime and not dependent on starting time of calculation. Note that the storage period needs to be connected with calendar periods, not only with number of days. The starting moisture at the beginning of the storage needs to be known. It can be defined by kiln method, typical in wood technology, but also by instant moisture meters (e.g. www.farmcomp.fi). In Finnish operational environment, the starting moisture can be set to 60% for fresh softwood and to 50% for fresh hardwood, if more accurate information is not available. Figure 4 illustrates the estimates created for three different piles established in different time of the year. When programming this algorithm into ICT-solutions, constraints for maximum and minimum moistures need to be used. For example, if starting the storing of fresh softwood (MC 60%) in early September, the moisture of wood does not increase up to 72% at the end of November, like it would go according to table 1. On the contrary, MC 25% wood pile stored from early April does not dry down to 8% before the end of July. Due to that, MC of 20% is appropriate minimum moisture allowed in Finnish calculations. In Finland constraints max MC 65% and min MC 20% are recommended. For other operational environments, constraints need to be defined by local experts. 5
Figure 4. Moisture estimates for three different piles established in October, March and June. Starting moisture 60%. 4 Discussion Drying algorithms based on daily moisture changes and drying periods have been originally developed in Sweden to forecast drying of pulpwood in roadside storages. The reason for developing algorithms for pulpwood was the same, saving in transportation costs per amount of dry matter. The approach based on drying periods and daily moisture change is suitable only when producing rough forecasts based on average situations. Especially the length of different drying periods can vary remarkably between years. When testing the algorithms, problems can occur especially in autumn periods where the difference between the predicted result and the expected value according to average weather statistics can be remarkable. Warm dry autumn can dry timber at the same time when algorithms estimates re-moisturing (September). When the trend is totally opposite than expected, the difference between estimation and reality can grow remarkably in relatively short time. The second generation of algorithms are already created in further studies. They are based on accurate weather statistics and the relation between net evaporation and drying of timber. Those algorithms require exact local weather monitor, which is not always possible. Algorithms presented in this paper can offer instant improvement for systems used for the management of energy wood procurement. Algorithms also enable calculations for 1 2 year period of tactical planning. This is required when energy companies are allocating their procurement operations in order to keep storages in appropriate quantity/quality relationship. 6
References Hillebrand, K. et.al. Polttopuun kuivaus ja laadun hallinta PUUT42, PUUY47. Julkaisussa: Alakangas E. (toim.). Puupolttoaineiden pientuotannon ja -käytön panostusalue, Vuosikatsaus 2005. Teknologiakatsaus 185/2005. Helsinki 2005. Röser, D., Mola-Yudego, B., Sikanen, L., Prinz, R., Gritten, D., Emer, B., Väätäinen, K. & Erkkilä, A. 2011. Natural drying treatments during seasonal storage of wood for bioenergy in different European locations. Biomass and Bioenergy. 35(10): 4238 (42)47. doi:10.1016/j.biombioe.2011.07.011 Nurmi, J. & Hillebrand, K. 2007. The characteristics of whole-tree fuel stocks from silvicultural cleanings and thinnings. Biomass & Bioenergy 31(6): 381 392. www.farmcomp.fi. Wile bio wood moisture meter for sawdust and wood pellets. Forest Energy Observer ISSN-L: 1799-5876 Forecasting Algorithm for Natural Drying of Energy Wood in Forest Storages Authors: Lauri Sikanen, Dominik Röser, Perttu Anttila, & Robert Prinz Publisher: COST Action FP0902, Metla Contact: Finnish Forest Research Institute (Metla) Yliopistokatu 6 80101 Joensuu Finland email: fp0902@metla.fi www: observer.forestenergy.org Editors: Lauri Sikanen, Dominik Röser, Robert Prinz Layout: Leena Karvinen, Robert Prinz 7