Climate implications of increased wood use in the construction sector - towards an integrated modeling framework

There is a growing interest in efficient ways to use biomass for the substitution of fossil fuels and non-biomass materials. Wood-based building material can affect the energy and carbon balances through at least four mechanisms: the relatively low fossil energy needed to manufacture wood products compared with alternative materials; the avoidance of industrial process carbon emissions; the increased availability of biofuels from biomass byproducts that can be used to replace fossil fuels; and the physical storage of carbon in wood building materials. Increased use of wood-based building materials will likely affect relative prices on timber markets. That translates into changed forest management, which in turn affects forest growth, biofuel availability and mitigation through carbon storage in the forest. A more comprehensive analysis of the climate effects of increased wood in the construction sector would be made possible by integrating a range of models. These include models of wood substitution, sector product markets, and forest management models on regional and stand levels. Several partial studies have been conducted in this field. Still, a modeling framework that extends all the way from the construction sector over international markets down to the individual forest stand has not been employed. The purpose of this study is twofold. One is the analysis of climatic implications of increased wood use in building construction. For this purpose, a new integrated modeling framework is developed (see Figure 1). This framework is then used for the analysis of four different wood construction scenarios. The other objective of the current pilot project is to demonstrate the viability of the proposed modeling approach and the improvements needed. Thus, it constitutes a preparation for more comprehensive future studies. The four wood construction scenarios depict wood consumption up to the year 2030 for the European construction sector. They are characterized as follows: • Base: “Business-as-usual” corresponding to a growth rate of total European softwood sawn wood consumption estimated at 1.48% annually up to 2030. (The results of the other scenarios are measured against the Base scenario.) • Sweden: Wood is used instead of conventional concrete construction for building apartment blocks in Europe, gradually increasing to 1 million flats per year by 2030. The construction data are from a case study of a building constructed in Växjö, Sweden. • Finland: The same as case Sweden, but using construction data from a case study of a building in Helsinki, Finland. • 1m3cap: Consumption of sawn wood in all European countries will reach 1.0 m3 per capita by 2030 (from the current average of 0.2 m3 per capita), corresponding to a European growth rate of 7% per year. This is a rather extreme and probably unrealistic scenario. Table 1 summarizes the effects at year 2030, i.e. the last year of the projection period. The emission reduction figures are based on computations for all materials composing the buildings. The emission balance takes account of (i) fossil fuel combustion for material processing and logistics, (ii) reduction of emissions due to replacing fossil fuel with biomass residues from harvest, processing and demolition, (iii) avoided emissions from cement process reactions, and (iv) carbon stock change in wood materials. The marginal energy is assumed to be coal. It should be noted that the impact on carbon stocks in the forests is not included in the emission balance; this effect is assessed by the forest models (see below). The roundwood demand in each year of the projection period was distributed among supplying countries by the EFI-GTM model. The EFI-GTM is a partial equilibrium model for the forest sector, i.e. it encompasses forestry, wood using industries, markets for round wood and forest industry products, and solves the market clearing problem by consumers' and producers' surplus maximization. The global market consists here of 55 regions, where almost all the European countries are represented by their own regions. The model contains markets for 34 forest sector commodities (25 forest industry products, five types of roundwood and chips, and four types of waste paper). For each region, supply functions for production factors are defined, as well as a set of fixed-input technologies with specific capacities for producing intermediate and final products. The forest supply is represented by supply elasticities for timber and pulpwood. The Swedish roundwood supply is described with an elasticity of 0.5 for both timber and pulpwood, based on runs with the Swedish regional forest model in this study. The model predicts that competition over the wood fibre increases in the future, and the prices of both pulpwood and saw logs rose in the Base case. Also, the Russian timber export tariff increases the prices in Scandinavia. In the Base case, the prices for softwood saw logs are projected to be 30% higher, and pulpwood prices 54% higher, in 2030 than in 2004. Note, however, that these increases can be considered a high estimate, since, for technical reasons, we chose to accept a higher rate of forest industry capacity accumulation than what we would have chosen in some other analyses. Table 2 shows the changes in Swedish harvest volumes and price changes in the scenarios Finland, Sweden and 1m3cap, compared to the Base case, in year 2030. The scenario Sweden had the lowest market impacts, with the softwood saw log price being about 3% up from the Base case level in 2030. In scenario 1m3cap, the saw log price more than doubled from the Base case due to the drastically increased softwood lumber demand. The growth in softwood lumber production made saw log chips supply abundant and led to a decline in the pulpwood harvests and price. In scenario 1m3cap, softwood pulpwood price came down by close to 20% compared to the Base. Due to the substitution effect in panel production, the hardwood pulpwood price also fell. Either harvest volumes or timber prices for Sweden could then be transferred from the EFI-GTM to the Swedish forest regional model. In this analysis prices were put in the forest regional model and the ensuing result studied. For the forest regional model for Sweden, the SMAC model was used. The model is an area matrix model where harvests are derived by assuming that forest owners maximize their net present value over an infinite horizon with current prices (i.e. they assume constant prices; solution procedure is value iteration on the Markov model). Since the SMAC model operates with 5-year growth periods, prices from EFI-GTM were averaged over 5-year periods. Comparisons between EFI-GTM and SMAC of harvest volumes were only conducted for Base and 1m3cap since price figures for scenarios Sweden and Finland are almost identical with the Base scenario. The saw log volumes of the two models are fairly similar for each of the scenarios over the first 10 years. However, for scenario 1m3cap during the rest of the period, where the EFI-GTM projects a substantial increase, the SMAC model presents a reduced harvest of saw logs despite a rather dramatic increase in saw log price. The reason for this reduction is essentially that it is profitable, with relatively more profitability in final felling compared with thinning, to postpone final harvests and increase both the relative and absolute yield of timber in the future. In the long run, after some 60 years, saw log supply became larger for 1m3cap than Base in the SMAC model projection. More detailed analyses of the management implications of the EFI-GTM price series were performed with a stand model, the Stand Management Assistant (SMA). SMA is an individual-tree, distance-independent growth and mortality model that finds optimal steady state stand management programs (planting density, timing and form of thinning and time of final harvest) by solving a non-linear, non-differentiable optimization problem with the Hooke and Jeeves method. The differences between scenarios Base and 1m3cap were studied for a range of stand types. In most cases the model predicted for 1m3cap: prolongation of the rotation period, increased number of thinnings, increased planting density, increased average standing volume, increased saw log production, reduction of pulpwood production, and increased average carbon stock. The effects were more pronounced on more fertile sites than in poorer sites, and more for spruce than for pine stands. The qualitative results from the SMAC and SMA models are more or less in agreement. Differences in the results can be attributed to the fact that stand establishment cannot change in the SMAC model, contrary to the SMA model, something which the SMA model shows has considerable influence on the design of the optimal management program. On the practical side the results indicate the following: • An increase of wood framed buildings would reduce net carbon emissions in the construction sector. The total net effect was not quantified because the changes could not be traced down to the forest. • The changes in the price relations between sawnwood and pulpwood of the EFI-GTM lead in the forest regional model to a change in the management programs towards prolonged rotations, leading to a medium term reduction of sawnwood supply. • The long term steady state analyses indicate small differences in carbon stock due to the price increases predicted by the scenarios. Sawnwood output increases, but is in most cases balanced by a similar reduction of pulpwood output. Rotations are prolonged and for several stand types the number of thinning is increased. The modeling system in this report represents an ambitious effort to combine models from different disciplines into one coherent system. It is no surprise that several gaps, overlaps and missing links have been detected. The following more general experiences were gained: • Linking the wood construction scenarios with the EFI-GTM, however demanding, works without major problems. The resulting demand for sawnwood can be distributed among countries by the EFI-GTM model in consistency with the construction scenarios. • The most problematic part of the system appears to be the linkage between the EFI-GTM and the forest regional model. In particular, the reaction of supply stemming from different price relations between sawnwood and pulpwood needs to harmonized. The SMAC model gives lower harvest volumes in Sweden in the first 2-3 decades compared to the EFI-GTM results because, seen from the forestry side, it is more profitable to postpone harvest given the assumed increase in saw log prices. To avoid this difference between EFI-GTM and the SMAC results one could run SMAC with both prices and volumes of saw logs and pulpwood fixed until 2030 according to the EFI-GTM results, so that only the forest management (silviculture and harvesting operations) are decided endogenously. It would also be advantageous to have the same temporal resolution in both models. • None of the models – the sector, regional or stand model – explicitly include biofuels. Given the growing importance of the biofuel market it would be desirable to adjust the models such that one could study the effects of changing demand and supply relations on economic indicators and forest management activities. • The detailed stand level model and the regional forest model could be better integrated with each other. • The overall consistency relies on a number of common parameters that are used in the different models, such as carbon emission factors and discount rates. They need keen attention to ensure consistency