Environ. Sci. Technol. 2009, 43, 6003–6009
Life Cycle Assessment of Oriented Strand Boards (OSB): from Process Innovation to Ecodesign E N R I C O B E N E T T O , * ,† M A R K O B E C K E R , ‡ ¨ LLE WELFRING† AND JOE CRP Henri Tudor/CRTE, 66 rue de Luxembourg, L-4002 Esch/Alzette, Luxembourg, and Kronospan Luxembourg S.A., BP 109, L-4902 Sanem, Luxembourg
Received September 18, 2008. Revised manuscript received June 10, 2009. Accepted June 18, 2009.
Oriented strand boards (OSBs) are wood panels that are used worldwide mainly in the packaging and the building sectors. Their market share is rapidly increasing thanks to their outstanding mechanical properties and to a renewed interest for wood based products. The OSB production process generates, nonetheless, emissions of volatile organic compounds (VOCs) during the air-drying of wood strands. This known problem in the literature leads to an odorous nuisance in the surrounding area of the production site. In order to address this problem, a novel application to wood drying of an innovative vapor drying technology is successfully operated at the production site of Kronospan Luxembourg S.A. In addition to the reduced odorous nuisance, a significant environmental added value is expected because of the other modifications induced by the vapordrying technique on the OSB production process viz. the reduced energy and raw materials demands and the change of adhesive mixture, with the addition of phenol resin. The potential impact of this technology on the OSB market is therefore very significant. This study was aimed at assessing the environmental added value provided by the vapor-drying technique through a life cycle assessment (LCA) according to ISO 14040-44 standards. The objective was to compare the environmental performances of the former and the current OSB production processes. Considering only the pollutant emissions from the OSB production process, the reduction of climate change impact and human health damage is significant, respectively, 15-20% and 50-75%. When the lifecycle processes related to the OSB production are included, the reduction of damages does not exceed 3-7%. Following an uncertainty analysis, this reduction was nevertheless proven to be statistically significant. However, it is observed that the reduction of environmental impacts and damages allowed by the vapordrying technology is counterbalanced by the change of adhesive mixture. Indeed the new adhesive mixture generates higher environmental damages than the former mixture because of the higher energy and raw material demand from phenol resin production. These results show the need to move from an approach focused on a single process innovation (the vapordrying technique) to a more general and systemic approach combining process and product ecodesign. Such approaches * Corresponding author phone: +352.42.59.91.603; fax: +352.42. 59.91.555; e-mail:
[email protected]. † CRP Henri Tudor/CRTE. ‡ Kronospan Luxembourg S.A. 10.1021/es900707u CCC: $40.75
Published on Web 07/13/2009
2009 American Chemical Society
are needed in order to effectively improve the overall environmental performance of a production system, without transfer of pollution along the lifecycle or offsets of pollution credits. LCA is definitively one of the most pertinent tools to identify improvement opportunities in the comparison of alternative designs from an environmental perspective.
1. Introduction Oriented strand boards (OSB) are wood based panels first produced in North America in 1965. They are built up in three layers of strands bonded with a resin and the strands of the outer layers are arranged at right angles to those in the middle layer (1). This structure leads to outstanding mechanical properties for application in several areas, mainly the building industry (e.g., for wall sheathing, roof panels, subfloors, single-layer floors, structural insulated panels, floor joists or rim boards), packaging, and the furniture sector (2). As a result, the market share of OSB is rapidly increasing. The OSBs dominate the North America structural panel marketplace, where they passed plywood production in 1999 (3). In Europe, the OSB production commenced in the early 1990s, and production capacity increased to over 3.6 million m3 in 2006. Kronospan Luxembourg S.A. (located in Sanem, Luxembourg) is the European leader in manufacturing high quality wood products and manufactures approximately 160 000 m3 of OSB panels per year. The OSB production at Kronospan entails seven steps (Supporting Information (SI) Figure S1): reception and storage of wood logs (“logging”); debarking and stranding of logs (“flaking”); drying of wood strands from 50 to 150% of moisture content to 5% (“drying”); screening of dry strands to separate the finest sizes (“screening”); blending of strands with adhesives mixture, wax and hardener (including sulfur, urea, and water), and forming in order to obtain a mat (“chemicals”, “blending+forming”); pressing of the mat to get the OSB (“pressing”). An additional final step of sanding and finishing could be required for some OSB products. The emission of volatile organic compounds (VOCs) during the drying of wood strands is a known problem in OSB production, as discussed in, e.g., refs 4, 5. These emissions vary according to the wood type and to the wood’s growing conditions. Terpenes are recognized to be the main VOCs generating odorous nuisances for the people living in the surroundings of an OSB production site. These substances are indeed mainly generated by the use of conventional airdrying techniques. The reason is that in this type of process only part (usually half) of the flow of exhaust gases from the dryer is recovered. The residual flow, with high terpene concentration, goes through a wet electrostatic precipitator before emission to the atmosphere. In order to reduce VOC emissions, a new drying process, called “ecodry”, has been successfully implemented since end 2004 at Kronospan Luxembourg S.A. The ecodry represents the first ever application worldwide of the vapordrying technique to wood drying. In this closed loop system, the output gases from the drying process are fully heated by the burner’s exhaust gases through a heat exchanger located outside the dryer. Part of the output flow from the heat exchanger is used to cool down the temperature of the exhaust gases from the burner, which allows feeding dry air to the dryer’s trammel as well. The inlet flow to the dryer trammel is therefore a vapor stream that enables the drying process. As a result, VOC emissions could be reduced because of the closed loop of the drying exhaust gases and the higher temperatures. VOL. 43, NO. 15, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6003
However, the influence of the ecodry on the OSB production process is likely to be even more important. Reduced fuel demand for heat production is expected since the ecodry systems recovers more heat from the drying exhaust gases. The ecodry system allows the burning of wood dusts instead of the exclusive burning of natural gas, which could further lower the demand of fossil fuels. Furthermore, the ecodry leads to modification of strands geometry and lowers their density. In the ecodry, the drying starts at the core of the wood strands and then moves to the surface, as opposed to the air-drying technique where it starts at the surface and moves to the core. The strand temperature is therefore higher than in the case of air-drying, which leads to a higher wood purity because of the distillation of terpenes and to the surface and density modification. The modification of the strands geometry requires, in turn, a change of the adhesive mixture. Before the implementation of the ecodry, the mixture included methyl urea phenol formaldehyde (MUPF), methyl urea formaldehyde (MUF), and polymeric methylene diphenyl diisocyanate (PMDI) resins. Using the ecodry technique, the required quantity of adhesive is lower than before (because of the lower strand density) but the MUF is replaced by MUPF, which provides higher resistance and maintains unchanged the OSB mechanical properties. Due to the reduced strand density, the amount of wood needed is reduced as well. As a result, the operation of the ecodry has a significant influence on the whole OSB production process lifecycle and involves significant technical challenges. The wide range of OSB applications and the current market trends make the expected impact of this process innovation in the OSB industry very important. In addition to the reduced VOC emissions and odorous nuisances, an important environmental added value is expected as well because of the lower energy and raw materials demands for OSB production. According to the environmental technology verification schemes (6) and other European policies (e.g., the European Technology Action Plan, ETAP), the evaluation of the environmental added value shall include the assessment of the lifecycle environmental impacts of the OSB production at Kronospan before and after the introduction of the ecodry. The aim of this study was to carry out this assessment, by collecting high quality field data related to the OSB production at Kronospan and considering two levels of analysis. At the “gate-to-gate” level, the reduction of the raw material and energy consumptions at the OSB production line, including, e.g., reduced pollutant emissions, odorous nuisances, and adhesive consumption, are quantified. This analysis highlights the environmental benefits generated directly onsite. Then, at the “lifecycle” level of analysis, the effects on the supply chain related to the OSB production, e.g., the electricity and natural gas production and delivery at the site, are included in order to have a comprehensive assessment of the OSB production. For both the levels of analysis, the life cycle assessment (LCA) methodology according to the ISO 14040-14044 standards (7) was considered. This article presents the results of the LCA and is intended to provide specific data and information for pertinent communication to the interested parties of the environmental performances of the ecodry technology. The environmental profile of the Kronospan’s OSBs products and the related improvement opportunities are illustrated as well. From a more general perspective, the results presented could support policy initiatives, e.g. the elaboration of criteria for an ecolabel on soft-floor coverings, including OSBs, which is indeed under development in Europe.
2. Materials and Methods 2.1. General LCA Methodology. The study was based on the current state of the art of the LCA practice (7-9) and on 6004
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 15, 2009
the ISO 14040-14044 standards (7). In particular, the ISO requirements concerning third party critical review for disclosure to the public of comparative assertions were fulfilled. An attributional LCA approach (9) was adopted, aiming at describing the environmental impacts attributed to the operation of the current OSB production system (with the ecodry technology) and to the former one (without the ecodry technology) at the Kronospan site. The indirect consequences generated by the introduction of the ecodry technology, which shall be considered in a consequential LCA approach (11), were not addressed in this study. These consequences are related to market oriented changes in the lifecycle processes related to the OSB production system following the introduction of the ecodry. As an example, the adoption of the new adhesive mixture in the OSB production could reduce the availability of some chemicals on the market and then contribute to the increase of production of substitutes. Despite the relevance to the aims of the study, consequential LCA is still debated and nonconsensual approach and, considered the challenges of the present study, the attributional approach was preferred. The software Umberto 5.0 (www.umberto.de) together with the Ecoinvent 1.3 database (www.ecoinvent.ch) was used for LCA calculations and collection of background data, e.g., concerning electricity and chemicals production. Production specific data were collected at the Kronospan’s site (see section 2.3). In order to have a comprehensive view of the lifecycle impacts of the OSB production systems, several lifecycle impact assessment (LCIA) methods were used. Cumulative energy demand (CED (12)) provides a valuable evaluation of the energy intensity of the OSB production. CED represents the overall energy consumed over the lifecycle and it is mainly related to the consumption of energetic raw materials. The assessment of climate change (midpoint) impact and endpoint damages using the Impact2002+ methodology (I2002, (13)) leads to a comprehensive view of the whole environmental burdens attributed to the studied systems. Endpoint damages on human health, resources, and ecosystem quality provide a consistent and concise view of the effects of the studied scenarios on the final targets. In order to evaluate the robustness of the LCIA results, endpoint damages have been assessed using an alternative methodology as well, Ecoindicator99 (EI99 (14)), which considers different assessment models and assumptions than I2002. Both the Ecoindicator99 and Impact 2002+ methodologies consider VOCs emissions lumped together, i.e., one damage characterization factor for the whole VOCs emissions. This is fully coherent with the inventory data available, which do not distinguish between VOCs species. The results of I2002 and EI99 have then been compared. The endpoint results were normalized and expressed as “points” which represent the number of EU inhabitant equivalent generating over one year the same damage than the studied system (15). The normalization factors differ between EI99 and I2002. 2.2. Functional Unit and Scope. The main function of the Kronospan’s OSB production process is the production of OSB sanded or unsanded, having a thickness ranging from 8 to 25 mm. Two additional functions have to be considered: the delivery of wood residues to be used in the production of medium density fiberboards (MDF) and the delivery of bark to be sold as a soil amendment. In this study, the functional unit was the production of 1 m3 of unsanded OSB/3 having a thickness of 15 mm. OSB type 3 has mechanical properties which make it suitable for use in load bearing in wet/moist conditions. The volumetric unit is used as reference to record production data onsite and was adopted in the Ecoinvent database and in previous studies (5). The seven steps of OSB production, described in the introduction (SI Figure S1), were included in the scope of the
study: logging, flaking, drying, screening, chemicals, blending and forming, and pressing. All the mass and energy flows involved in these steps were included in the assessment. The production of the process infrastructure was excluded from the scope of this study because the infrastructure was not significantly changed by the introduction of the ecodry. Since the background data are issued from the Ecoinvent v1.3 data sets, the boundaries of the scope of the study are implicitly set by the cutoff rules established within Ecoinvent. Based on this OSB production system, four scenarios were considered to compare the former and current OSB production process: “2004b” represents the OSB production before the ecodry; “2004a”, “2005”, and “2006” represent the OSB production after the ecodry, following the respective yearly production data. Since the ecodry was introduced toward the end 2004, the data for 2004a are extrapolated data as if the process had been operated all the year with the ecodry. These scenarios are fully comparable because the former and current OSBs target the same market and fulfill the same functions. The OSB manufacturing is indeed very complex and many assumptions and parameters are considered throughout the study. The process has to be continuously adapted because of the nonhomogeneity of wood parameters (mainly density and moisture content), which depend on the type of tree (pine, douglas, epicea), on the area where the tree grows and on the season. For this reason, and in accordance with the ISO standard, sensitivity analysis was considered before the interpretation of the LCA results (see section 3.3). 2.3. Data Collection and Data Quality Analysis. The collection of high quality primary data related to the OSB production at Kronospan is primordial for the consistency and representativeness of the LCA results. Foreground data were issued from the Kronospan’s recordings, which includes sampled concentrations and hourly flows of pollutant emissions and off-gas volume flows; operational conditions for dryer and press; monthly average production and raw material consumption data. Pollutant emissions data have been systematically checked using the concentrations and off-gas volume flows. VOCs emissions were lumped together and measured as “total VOC” and “condensable VOC”, the latter including only the condensable part of emissions. An analysis of variance (ANOVA) with F-test was performed in order to evaluate if the different samples belong to the same population, i.e., the average values can be used in the calculation. Monthly production data were screened in order to skip wrong recordings. The quality of these primary data has been reviewed by defining data quality criteria and by setting quality requirements (bad, average, good) for each criterion according to the ISO14044 standard (SI Table S1). A quality level “good” is required for the criteria time related coverage, geographical correlation, and technological correlation. The other criteria are generally more difficult to address and could not be fulfilled at a quality level higher than “moderate”. The latter is especially true for representativeness, which is related to the number of repeated measurements and recordings available, which are rarely sufficient in the practice. The evaluation of the quality of the life cycle inventory (LCI) has been carried out using the methodology provided in Umberto 5.0, which is adapted from ref 16. Quality levels for each data quality category were defined (SI Table S2) and then each inventory data was evaluated by choosing a quality level (from bad to good) based on the information available. The data quality of the whole LCI was finally assessed by aggregating the quality levels of all the inventory data, by choosing one of the aggregation methods proposed by the Umberto software. The final result (SI Table S3) appears to be highly dependent upon this choice but, nonetheless, some conclusions can be drawn. As expected, the main limiting factor is the repre-
FIGURE 1. Odorous emissions from wood drying (averages from measurements and standard deviation). The suffixes “a” and “b” means “after” and “before” the implementation of the ecodry, respectively. sentativeness of the recordings, especially for pollutant emissions. Production data (e.g., natural gas consumption, OSB production rates) are more representative since monthly averages are available. However, bad or average aggregated data quality appears to be due to specific measurements on a specific criterion. For this reason, the data quality is considered to be satisfactory to address the aims of the present study. 2.4. Multifunctionality. As already described, the OSB production system fulfils two cofunctions: the delivery of wood screening residues to be used as additional fuel in MDF production and the delivery of bark to be sold as a soil amendment. Therefore, part of the LCI should be attributed to these cofunctions and to not be considered in the inventory of the OSB production. Otherwise, the OSB production would be charged of environmental loads which were actually not pertinent. The way to address this multifunctionality problem, i.e., to determine the part of the LCI to be attributed to the cofunctions, depends on the LCA approach (attributional or consequential). Several methodological approaches are debated in the LCA community and no unique solution exist (8, 9, 17). Since this study focuses on attributional LCA, a possible solution to multifunctionality is allocation. In the case of wood screening residues, data were allocated according to the mass of the residues and the mass of the wood included in the OSB. This corresponds to a “process subdivision approach” according to the ISO 14040 terminology. As a result, the quantity of raw wood needed for the OSB production is 613.3 kg/m3 OSB (92%), whereas the amount of raw wood leading to residues is 50.7 kg/m3 OSB (8%). In the case of bark, the default approach was to allocate the LCI data on the basis of the market price of the products: 30 euros/t for bark and 280 euros/t for the OSB. These are price estimates, which may be subject to change. However, regardless of the LCA scenario, the allocation ratio for OSB is always higher than 99%. This means that the energy and material flows of the lifecycle of logging and flaking, which are the processes contributing to the production of bark, were almost entirely allocated to the OSB. In the sensitivity analysis, the process subdivision approach was studied as well (see section 3.3).
3. Results 3.1. Gate to Gate Impact Assessment. The ecodry technique is primarily aimed at reducing odorous emissions from wood drying and at lowering energy demand. Odorous emissions have been measured by an agreed organism using a standardized method (18) and are expressed as GE/m3, where GE stands for “odor” in German. As a result of the ecodry, odorous emissions have decreased by 30% (Figure 1) and VOL. 43, NO. 15, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6005
FIGURE 2. Normalized end point damages (I2002) at the gate-to-gate level. The suffixes “a” and “b” means “after” and “before” the implementation of the ecodry, respectively.
FIGURE 3. End point damages (I2002) at the lifecycle level. The suffixes “a” and “b” means “after” and “before” the implementation of the ecodry, respectively.
the total energy demand has decreased by 10-15% (SI Figure S2) after the introduction of the ecodry. These improvements led to a reduction of the environmental impacts and damages as well (Figure 2). According to the I2002 methodology, the normalized contribution to the climate change impact is decreased by 15-20%, mainly as a result of the lower CO2 emissions from the drying and the chemicals production phases. The contribution to human health damage is decreased by 50-75%, because of the lower VOCs and particulate emissions (SI Table S4). However, NOx emissions (including NO2 emissions from the dryer) have become the main contributor to human health in the current OSB production process, using the ecodry technique. All these results can be explained by considering the reduced availability of wood residues as energy source for drying. Before the implementation of the ecodry, energy was mostly provided by wood residues, which generate mainly particulate emissions during combustion. After the installation of the ecodry, natural gas has become the main energy source, producing thermal NOx emissions. Since the overall energy demand at the dryer has decreased following the introduction of the ecodry technique, the increased ratio of natural gas over wood residues (which would have occurred independently from the ecodry) highlights even more clearly the added value provided by this technology. Ecosystem quality damage remains unchanged (Figure 2). However, the normalized damage magnitude is far lower than for climate change and human health. Damage on resources cannot be calculated at the “gate-to-gate” level of analysis because no mineral of fossil resource are directly consumed. This damage is addressed at the “lifecycle” level of analysis. 3.2. Life Cycle Impact Assessment (LCIA). At the lifecycle level, several LCIA methodologies were combined in order to have a comprehensive view of the environmental impacts and damages generated. The cumulative energy demand (CED) of OSB production is driven by biomass and fossil resources, i.e., mainly wood and natural gas consumption (SI Figure S3). The adoption of the ecodry does not significantly reduce the wood consumption and thus the CED reduction is limited to 3%. The climate change impact was evaluated at two different time horizons: 500a, as in I2002 and 100a, which is considered in EI99 and more commonly used in policy making (SI Figure S4). The reason to use two different time horizons is to analyze the relative contribution to climate change of substances which have a lifetime between 100a and 500a. The results are relatively similar which means that the main contributor to climate change is CO2. According to the normalized end point damage results from I2002 (Figure 3), the contribution to climate change is more than four times higher than at the gate-to-gate level (Figure 2) and is decreased by 5-8% by the adoption of the ecodry. The percentage decrease depends on the scenario considered.
Damages on resources and on ecosystem quality are nearly unchanged by the ecodry (Figure 3). The damage on human health damage is reduced by approximately 10%. However, the magnitude of the normalized damage on resources is higher than the damage on human health. Thus it is difficult to evaluate whether the ecodry has provided a clear advantage or not following the 10% reduction of human health damage. The damages on resources, human health and ecosystem quality were therefore aggregated into a single score damage. At this extent, a weighting scheme should be defined through the elicitation of decision makers’ preferences. This type of weighting could not be elaborated in the framework of this study. As first estimate, the weighting factors (human health: 40%; ecosystem quality: 40%; resources: 20%) defined in ref 14 were used. The single score damage reduction is between 5 and 6%, which is similar to the result obtained for climate change (SI Figure S5). In the case of higher weighting of the damage on resources, the reduction would be lower and therefore the environmental performances of the former and current OSB processes could no longer be distinguished. The production of chemicals (mainly adhesives), the drying process, the electricity production and the wood transport are the main contributors to the climate change impact (respectively 43, 25, 13, and 10% in the scenario 2006) and to the damages (SI Table S5). The main substances concerned are CO2 emissions (92% contribution to climate change), particulates and NOx emissions (respectively, 25 and 47% contribution to human health damage) and natural gas and crude oil resources (respectively, 52 and 30% contribution to damage on resources). All these substances are related to combustion processes in the lifecycle (SI Table S6). Based on results in Figure 5 and SI Table S6, we note that the contribution of the chemicals to the single score damage systematically increases, both in percentage and in absolute value. This is mainly due to the increase of the MUPF consumption because of the change of adhesive mixture with the ecodry. Based on the inventory data from ref 19, the MUPF production process has higher energy and raw materials demand than the MUF production and thus leads to higher damages. This effect partly offsets the damage reduction obtained because of the reduced consumption of wood, electricity and natural gas, as already described. The consideration of EI99, as alternative LCIA methodology, basically confirms the LCIA results obtained using I2002, with a single score damage reduction between 2 and 3% (SI Figure S6). As already mentioned, in the EI99 single score damage, climate change (assessed at the 100a time horizon) is included by assessing its contribution to human health damage, despite significant modeling uncertainties (Goedkoop and Spriensma, 2000). Thus, the single score damage allows comparing the OSB production processes by considering all the environmental effects altogether, which is not the case of I2002. The only significant difference observed
6006
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 15, 2009
TABLE 1. Sensitivity Analysis Cases
case DW280
description 3
parameter concerned 3
wood density can vary from 280 kg/m (spruce) to 450 kg/m (pine)
value set
default value considered
wood density
280
450
MC50DW280 wood moisture content (MC) can vary from 50% to 150%. MC and wood density are interrelated. MC50DW450
MC density MC density
50 450 50 280
140 450 140 450
CVOC
VOC emissions at dryer can be expressed as “total VOC” or “condensable VOC” (which are part of “total VOC”)
VOC
condensable VOC total VOC
electricity
splitting up of the whole electricity consumption recorded between the dryer (RXD), the flakers (RXF) and the press (RXP) processes
RXF RXD RXP
30% 50% 20%
30% 40% 30%
allocation
allocation criteria
allocation
mass
economic
concerns the magnitude of ecosystem quality damage, which is higher for EI99 than for I2002 because of the higher consideration of land occupation, as already noticed by, e.g., ref 20. Land occupation is related to the phase of wood production, which indeed has a much higher contribution than in I2002 and probably is overestimated to some extent. Based on these results, it is recommended to concentrate the improvement efforts mainly on the adhesive mixture and on the operation of the drying process, focusing on the substances which have been highlighted. The efforts are nevertheless depended on the degree of influence of the OSB manufacturer on these phases and substances. The degree of influence on the choice of chemicals is rather high since the adhesive mixture could be further modified upon the testing of mechanical properties of the OSB produced. The influence on the drying process is low because major improvements have already been obtained through the implementation of the ecodry. 3.3. Sensitivity and Uncertainty Analysis. The LCA scenarios studied involve choices, assumptions, and process parameters variations, which could significantly affect the results of the study. A first screening of the LCA model allowed the identification of a few parameters, listed in Table 1, which have a significant influence on the LCI. First, the wood density and the moisture content define the mass of wood to be dried, which is a key issue in calculations. Then, VOCs emissions can be calculated by considering only the condensable part or the total emissions, which could make a significant difference in terms of human health damage. Regarding the electricity consumption, the recordings concern the overall OSB production line. As a result, the consumption has to be split between the production phases. The part of electricity attributed to the flaking and the drying is not fully considered because of the allocations (multifunctionality). Therefore, depending on the split, the amount of electricity consumption attributed to the OSB process could vary significantly. Finally, the choice of the allocation criteria has a direct influence on the inventory data attributed to the OSB production. Six main cases of sensitivity analysis were considered, with a combination of realistic variations of these parameters (Table 1). After recalculation of the LCIA results for all the cases, it is shown that the variations of wood density and moisture content can increase the LCIA results up to 10% (Figure 4). This variation does not modify, however, the comparison between the scenarios because it concerns parameters which are not modified by the introduction of the ecodry and therefore it applies to the same extent to all the scenarios. Because of the limited number of parameters and likely values considered, the sensitivity analysis covers indeed only part of the uncertainties affecting the LCIA results. The
sensitivity analysis gives only a first indication about the possible range of variability of the results. In order to check the consistency of the LCA results, a comprehensive uncertainty analysis is required by ISO standards. According to the current state of the art, a Monte Carlo approach was adopted in this study. The uncertainty on primary data (mainly related to Kronospan’s production process) was characterized by means of uncertainty distributions, derived from data recordings and data collection. The distributions were specified through the parameters considered in the Umberto model (SI Table S7). The average and the standard deviation of the distributions were calculated from recordings. The shape of distributions was assumed to be of the “normal” type, which is the most commonly used in LCA. The LCI and LCIA calculations have been repeated hundreds of times. At each iteration, a random value of each inventory data was chosen according to the uncertainty distribution shape. Calculations were performed within Umberto 5.0, using native interfaces and algorithms, and taking into account the correlations between data sets, in order to avoid the risk of overestimating the uncertainty. This risk basically concerns the data which are considered by all the scenarios. At each iteration, the same sampled random values of these data have to be considered in all the scenarios otherwise the scattering on sampling would add nonrelevant uncertainty to the final LCIA distributions. Once the iterations have been completed, statistical distributions of the LCIA results were obtained. The distributions obtained for the damage on resources and climate change impact of the scenarios 2004b and 2006 are illustrated in SI Figure S7 and Figure 5, respectively. The distributions overlap and therefore statistical test (t test) was used in order to evaluate the significance
FIGURE 4. Sensitivity analysis results: range of variation of LCIA results (solid line) as compared to the baseline case, and averages of LCIA results (marker). Please refer to Table 1 for the explanation of the different cases. VOL. 43, NO. 15, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6007
FIGURE 5. Uncertainty distributions of climate change impact from Monte Carlo analysis. The suffix “b” means “before” the implementation of the ecodry. Please note that climate change impact results are not normalized.
of wood transport is proven be important. Depending on the technical constrains and the availability of wood, it is recommended to prefer low moisture content and highdensity wood, which could dramatically reduce the share of damages due to transport. All the above-mentioned aspects should be taken into account in the definition of ecolabeling schemes for OSB products and should be of highest priority in the analysis and improvement of production processes. From a more general perspective, the results of this study show the need to move from a single process innovation approach, e.g., the ecodry technology, to a more general and systemic approach combining process and product ecodesign. Such approaches are needed in order to effectively improve the overall environmental performance of a production system, without transfer of pollution along the lifecycle or offsets of pollution credits. LCA is definitively one of the most pertinent tools to identify improvement opportunities in the comparison of alternative designs from an environmental perspective.
Acknowledgments of the differences between the averages. A preliminary test for the equality of variances (F-test) indicated that the variances of the results for climate change and resources are significantly different with respectively p ) 0.007 and p ) 0.002 (p , 0.05, which is the confidence interval). Therefore, a two-sample t test assuming equal variances was performed, which confirmed that for both the damages, the difference is statistically significant, especially for resources (p ) 1 × 10-18 , 0.05). As a result, sensitivity and uncertainty analysis confirmed that the lower environmental damages and impact of scenario 2006, as compared to 2004b, are robust with respect to significant parameters variations and statistically significant according to t test. The introduction of the ecodry has definitively improved the environmental performance of the OSB production process.
4. Discussion The introduction of the ecodry technology for wood strand drying in the OSB production process was expected to reduce the VOC (odorous) emissions on site and to provide a significant environmental added value, in terms of reduced contributions to environmental impacts and damages. The LCA study has confirmed these advantages. Nevertheless, by using a lifecycle approach, the study has showed that the reduction of climate change impact and environmental damages (or single score damage) does not exceed 5-8%. These results are nonetheless statistically significant. The reduction could have been more significant if the environmental criteria had been considered in the choice of the adhesive mixture. Indeed, the potential reduction of environmental impacts, resulting mainly from the reduced heat demand and from the lower drying emissions, is less important than expected because of the higher impacts associated to the new adhesive mixture and particularly to MUPF resins. In order to further reduce the environmental pressure of OSB production, we suggest to test and to assess new adhesive types and mixtures, with a special focus on the replacement of MUPF. As an example, recent experiences show that the use of kraft lignin phenolic resins could be a promising alternative (21). Furthermore, we recommend increasing the part of renewable energy used for drying by using, upon availability, more wood residues. These modifications would allow reducing the damage on resources (i.e., depletion of natural gas resources) and human health (i.e., NOx emissions from drying). Special attention has to be paid to a potential increase of particulate emissions, which would generate higher damage on human health. The contribution 6008
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 15, 2009
We acknowledge the funding of the project by the LIFE Environment scheme of the European Community (ECOSB project, LIFE05/ENV/L/000047). The ECOSB project is among the five LIFE Environment projects selected as the “Best of the Best” 2008-2009. The comments received from four anonymous reviewers are also gratefully acknowledged.
Supporting Information Available Table S1: Data quality criteria and requirements. Table S2: Data quality categories and levels. Table S3: Data quality assessment results. Table S4: Analysis of the contribution of processes and substances at the gate-to-gate level. Table S5: Analysis of the contribution of processes at the lifecycle level. Table S6: Analysis of the contribution of substances at the lifecycle level. Table S7: Parameters considered in uncertainty analysis and related Normal distributions. Figure S1: OSB production system modeled in Umberto 5.0. Figure S2: Heat demand at dryer (averages with standard deviation calculated from Kronospan’s recordings). Figure S3: Cumulative energy demand (CED) at the lifecycle level. Figure S4: Climate change impact (100a and 500a time horizon). Figure S5: Single score end point damage (I2002) at the lifecycle level. Figure S6: Single score end point damage (EI99) at the lifecycle level. Figure S7: Uncertainty distributions of damage on resources from Monte Carlo analysis.This material is available free of charge via the Internet at http://pubs.acs.org.
Literature Cited (1) Oriented Strand Boards (OSB). Definitions, Classification and Specifications, EN 300-97; CEN, 1997. (2) Rebollar, M.; Pe´rez, R.; Vidal, R. Comparison between oriented strand boards and other wood-based panels for the manufacture of furniture. Mater. Des. 2007, 28, 882–888. (3) Hansen, H. Structural panel industry evolution: Implications for innovation and new product development. For. Policy Econ. 2006, 8, 774–783. (4) Manninen, A. M.; Pertti, P.; Holopainen, J. K. Comparing the VOC emissions between air-dried and heat-dried Scots pine wood. Atmos. Environ. 2002, 36, 1763–1768. (5) Kline, D. E. Gate-to-gate life cycle inventory of oriented strandboard production. Wood Fiber Sci. 2005, 37, 74–84. (6) Calleja, I.; Delgado, L. European environmental technology plan (ETAP). J. Clean. Prod. 2008, S181–S183. (7) Environmental ManagementsLife Cycle Assessment ISO 14040 Principles and Framework ISO 14044 Requirements and Guidelines; ISO-International Organisation For Standardisation: Geneva, 2006. (8) The International Reference Life Cycle Data System (ILCD) and the ELCD Core Database; European Commission JRC2007; http://lca.jrc.ec.europa.eu.
(9) Lundie, S.; Ciroth, A.; Huppes, G. Inventory Methods in LCA: Towards Consistency and Improvement. Final Report, UNEPSETAC Life Cycle Initiative Life Cycle Inventory (LCI) Programme Taskforce 3: Methodological Consistency; UNEP-SETAC: Geneva, 2007. (10) Zamagni A., Buttol P., Porta P., Buonamici R., Masoni P., Guine´e J., Heijungs R., Ekvall T., Bersani R., Bienkovska A., Pretato U. Critical Review of the Current Research Needs and Limitations Related to ISO-LCA Practice. Deliverable of the WP5 of the FP6 Coordination Action for Innovation in Life Cycle Assessment (CALCAS); ENEA: Bologna, 2008. (11) Weidema, B. P. Geographical, Technological and Temporal Delimitation in LCA, Technical guidelines for product life cycle assessment no. 3; The Danish Environmental Protection Agency: Copenhagen, 2003. (12) Frischknecht, R.; Jungbluth, N.; Althaus, H. J.; Doka, G.; Heck, T.; Hellweg, S.; Hischier, R.; Nemecek, T.; Rebitzer, G.; Spielmann, M. Overview and Methodology, Ecoinvent Report No. 1; Swiss Centre for Life Cycle Inventories: Du ¨ bendorf, 2004. (13) Jolliet, O.; Margni, M.; Charles, R.; Humbert, S.; Payet, J.; Rebitzer, G.; Rosenbaum, R. IMPACT 2002+: A new life cycle impact assessment methodology. Int. J. Life Cycle Assess. 2003, 8 (6), 324–330. (14) Goedkoop, M.; Spriensma, R. The Eco-indicator 99: A Damage Oriented Method for Life Cycle Assessment. Methodology Report, 2nd ed.; Pre´ Consultants: Amersfoort (NL): The Netherlands, 2000.
(15) Humbert, S., Margni, M., Jolliet, O. IMPACT 2002+: User Guide. Draft for Version 2.1; Lausanne: EPFL, 2005. (16) Weidema, B.; Wesnoes, M. S. Data Quality Management for Life-Cycle Inventoriessan Example of Using Data Quality Indicators. Invitational Expert Seminar on Uncertainty and Statistics for Product Life Cycle Assessment; Technical University of Delft: Delft, The Netherlands, 1995. (17) Rebitzer, G.; Ekvall, T.; Frischknecht, R.; et al. Life cycle assessment Part 1: Framework, goal and scope definition, inventory analysis, and applications. Environ. Int. 2004, 30, 701– 720. (18) University of Stuttgart. Olfaktometrie (Geruchsmessung). http:// www.iswa.uni-stuttgart.de/sia/analytik/a_o.htm. Accessed July, 22th, 2008. (19) Althaus H.-J., Chudacoff M., Hischier R., Jungbluth N., Osses M. Primas A. Life Cycle Inventories of Chemicals, Ecoinvent Report No. 8, v2.0; Swiss Centre for Life Cycle Inventories: Du ¨ bendorf, 2007. (20) Zah, R.; Bo¨ni, H.; Gauch, M.; Hischier, R.; Lehmann, M.; Wa¨ger, ¨ kobilanzen von Energieprodukten: o¨kologische Bewertung P. O von Biotreibstoffen; Swiss Federal Office of Energy: Bern, 2007. (21) Cavdar, A. D.; Kalaycioglu, H.; Hiziroglu, S. Some of the properties of oriented strandboard manufactured using kraft lignin phenolic resin. J. Mater. Process. Technol. 2008, 202, 559563.
ES900707U
VOL. 43, NO. 15, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
6009
