The Rijksmuseum Amsterdam - Hygrothermal analysis and dimensioning of thermal insulation

The Rijksmuseum Amsterdam - Hygrothermal analysis and dimensioning of thermal insulation J. Grunewald, U. Ruisinger & P. Häupl Department of Mechanical and Aerospace Engineering, Syracuse University, 9 Link Hall, Syracuse, NY 3-, USA Dresden University of Technology, Institute of Building Climatology, Zellescher Weg 7, D-6 Dresden, Germany Contact: john.grunewald@gmx.de ABSTRACT: The vitalization of the historical building stock demands planning of reconstruction measures that increasingly aim at protection of our cultural heritage and at energy-efficient renovation as well. But these requirements can be contradictory sometimes. Questions arise, e.g.: What are the potentials and the risks of a thermal insulation? Which materials and which thickness would be appropriate for application in historical buildings? The external walls of the Rijksmuseum Amsterdam, being under reconstruction currently, suffered from moisture originated damages. Therefore, it was intended to apply an adequate thermal insulation to provide a proper micro climate near the internal wall surfaces and to minimize fluctuations in indoor climate. The insulation has to be properly dimensioned to limit extreme values in temperature and relative humidity. In addition, the measures should reduce interstitial condensation and prevent consecutive frost damages at outer parts of the facade. Two internal insulation options were discussed, one based on vapor-tight cellular glass and another one using capillary-active calcium silicate. Both variants and, as a reference case, the existing construction without insulation were investigated by application of numerical simulation software. Results are presented here for undisturbed one-dimensional wall constructions exposed to climate conditions of Amsterdam on an hourly basis. INTRODUCTION The protection of historical buildings often demands measures due to their particular utilization requirements. Examples are libraries or museums where indoor temperature and relative humidity have to be kept within certain tolerances in order to maintain exhibits and works of art permanently in a good state. A thermal upgrade of the building envelope by external insulation is often not applicable to historical facades. Aspects of energy-efficient renovation supervene in old residential houses. These contradictory requirements put questions about potentials and the risks of internal thermal insulation. The Rijksmuseum Amsterdam, as one of the ten principal museums in the world, is widely acclaimed for its collection of paintings from the Golden Age. The Rijksmuseum is a museum of art and history with five different collections under a single roof painting, sculpture and decorative arts, drawings, prints and photographs, Asiatic art, and Dutch history. The Rijksmuseum has numerous features that no longer meet today s safety requirements, and modern concepts regarding many aspects make an urgent call for new facilities. The starting point of the renovation is to restore light, air and architectural transparency to the building, and to design it so that it meets the needs of today s visitors. The renovation is one of the most radical operations the Rijksmuseum has ever had to undergo. Figure. The Rijksmuseum in the city of Amsterdam is a cultural centre of excellent reputation. As part of the planning works, an expertise was prepared by Arup (), stating that among other building physical problems the external walls show abnormal high moisture contents. Gravimetric measurements of selected bricks taken from different

parts of the building envelope ware carried out in the Building Physical Laboratory of the Institute of Building Climatology. Higher moisture penetration, sometimes full saturation, was found. Since rising damp can be excluded as reason for the high moisture contents, condensation and rain penetration were the remaining possible reasons. During the rehabilitation works, the facade was checked for cracks and open joints and, if necessary, repaired. Most windows have been replaced, which was considered a reasonable measure for safety, art conservation and reduction of heating energy losses. Naturally, due to the lower thermal transmittance of the new windows, the sheet glass surface temperature is higher than before. Therefore, substitution of the old windows had to be accompanied by a thermal upgrade of thermally weak parts of the outer walls. If the adjacent wall constructions are not insulated additionally, there would be a risk of condensation at the wall surfaces. Even if no condensation on the inner surface of outer walls takes place, there is a risk for mould growth. The critical relative humidity for mould growth is disputed among experts. Under optimal conditions - depending, for example, on the period of the climatic load, temperature and organic substrate - very few mould species grow on surfaces buildings below 7% relative humidity, Sedlbauer (). It is undisputed that a relative humidity higher than 8% should be avoided in any case. In this expertise it is considered acceptable if a relative humidity of 7% is not exceeded for more than - partially sequentially - ten days per year, Seifert et al. (). As one conclusion of the current situation, it was decided to provide an adequate internal thermal insulation of the exterior facades. Two insulation options were discussed: 3mm cellular glass, attached with mm glue mortar, and with a mm clay plaster layer as surface finish on the interior, calcium silicate, thickness still to be determined, attached with mm glue mortar, and with a mm lime cement plaster layer as surface finish on the interior. The Institute of Building Climatology was instructed to prepare several expertises for decision making. The hygrothermal behavior of the two proposed insulation options, in comparison to the non-insulated case were investigated. In addition, construction details as window recesses, air ducts and geometric thermal bridges were examined. As result of the hygrothermal analysis, recommendations for the constructive execution were given. The analysis was carried out using the simulation software DELPHIN developed at the Institute for Building Climatology at the Dresden University of Technology. CLIMATIC CONDITIONS During the simulation, the constructions are exposed to non-steady-state climatic conditions at interior and exterior sides. The annual course of temperature and relative humidity, the impact of driving rain, short and long wave radiation were taken into account on hourly basis. An hourly climatic data set of Amsterdam provided by Arup () was used for the external conditions; the set contained temperature, relative humidity, radiation and wind measured over one year. Hourly rainfall values were taken from the weather station most close to the city of Amsterdam. The monthly rainfall values (in total 88 l/m² a) used in the analysis are slightly higher than the monthly average rainfall data of Amsterdam from 98 to 99 (in total 83 l/m² a). The driving rain density on the wall surfaces is calculated from wind direction, wind speed and rain fall density on a horizontal surface area (Figure ). The highest rainfall density is reached at constructions exposed to the west, here, about / of the rain fall density on a horizontal surface area. That is reason why, in order to be on the save side, all build-ups have been analyzed as west-facing. Rain fall density in l/mh 7 6 3 Time in days Figure. Rainfall on a horizontal surface (data from nordic coast region of Germany) The impact of direct solar radiation has been calculated for a west-facing wall too. In addition, the diffuse solar radiation and the long wave radiation balance were taken into account. The simulations are usually conducted for a time period of several years. It was assumed that the climatic conditions will recur periodically every year. Interior conditions are described according to the ambient design conditions communicated by Arup () (summer 3 C / % r.h., winter C / % r.h., tolerance ±K / % r.h.). The interior climate has been modelled by sinusoidal functions as shown in Figure 3. The temperature curve represents exactly the proposed conditions, without the allowed tolerance of K. To achieve conservative, worst-case results, the relative humidity is assumed to be at the upper limit of its tolerance (%) band, which could be caused, for example, by a high visitor frequency. 3 3

To achieve results of two-dimensional simulations in a reasonable time, some details of the construction have been simplified. Because the software can reproduce rectangular edges only, curved surfaces have been modelled as even. Double glass panes have been simulated as one pane with adjusted values identical to double panes. Coatings on the internal surface has been assumed diffusion-open. The effects on results are assumed to be very small. The simplifications were chosen such that the results will always be on the safer side, i.e. the real hygrothermal behavior will be less problematic. Since the time for preparation of the expertises was short for material measurements, most of hygrothermal material data has been taken from the material data base of DELPHIN. The material parameters of brick incorporate vapor diffusion resistance values and moisture retention values from Dutch guidelines for building materials too. At the same time, it was agreed on conduction of supplementary material tests to be carried out at the Institute of Building Climatology and on subsequent simulations with updated material data to check the results. The most important material parameters used for simulations are collected in Table. The two insulation materials have oppositional properties. Cellular glass has no liquid water conductivity and a very high resistance to water vapor diffusion, whereas calcium silicate is characterized by high liquid water conductivity (capillary activity) and low water vapor diffusion resistance. The material parameters presented in Table form the basic input to the simulation program. The storage and transport coefficients used in differential equations for heat and moisture transfer require specification of material properties as functions of temperature and moisture content. To generate those functions, material models are employed. Here the material property specification is based on the Engineering model of hygrothermal material characteristics Scheffler & Grunewald (). 3 Table. Basic material parameters for simulation Temperature in C - - 9 8 7 6 3 Indoor temperature (ARUP-specification) Outdoor temperature Time in days Time in days 3 Indoor relative humidity (ARUP-specification) Outdoor relative humidity 3 3 3 Density Specific heat capacity Thermal conductivity Vapour diffusion resistance Water content at 8% R.H. Capillary saturation Water uptake coefficient / kgk / Symbol ρ c λ R µ dry θ 8 θ cap A w Unit kg / J W - m 3 /m 3 m 3 /m 3 kg / m 3 mk m s. Historical 7 8.8 9.6.3. brick LC-plaster 8.... Cellular 7. 7 e-. glass Glue mortar 6.7 3.73.3. Clay plaster 7.87..3. Calcium 7.6.6.8. silicate Clinker 9.96.8.8.3 Figure 3. Courses of indoor and outdoor temperature and relative humidity used for simulation 3 MATERIAL DATA CONSTRUCTION Two insulation options were investigated, cellular glass and calcium silicate, in comparison to the existing non-insulated wall construction. The three buildups used for analysis of the undisturbed wall area are represented in Figure. Layering, dimensions and material selection followed the specifications made by Arup (). The numerical analysis includes: annual courses of temperature, relative humidity and water contents, temporal courses of integral moisture contents and condensate mass, proof of condensation risk at internal wall surfaces as well as compliance of humidity criterion concerning mold growth (short term exceeding of 7% R.H. for maximum days), quantification of energy savings. The initial conditions were chosen to account for the experimental results of the brick moisture mea-

surements. According to own investigations, the inner brick range (historical brick) was assumed entirely saturated by about Vol%. The outer clinker layer and the insulation layer were set initially to moisture contents in equilibrium to 8% R.H. The initial temperature of the whole construction was C. Without insulation Cellular glass insulation mm Clay plaster, 3mm Cellular glass, mm Glue mortar Calcium silicate insulation mm Lime cement plaster, 3mm Calcium silicate, mm Glue mortar Figure. Build-up of the existing non-insulated construction (top), insulation option one with cellular glass (middle) and insulation option two with calcium silicate (below) After a certain settling time, the physical quantities (e.g. water content and temperature) in every part of the construction recur in the simulation over the year. This means that they reach, according to the cyclic climatic conditions, a quasi-stationary state. This process can take years, but the main changes take place during the first months or years. This process is pronounced by the material properties and by the chosen initial conditions. The duration of simulations of five years was long enough to reach almost quasi-stationary conditions. RESULTS mm mm 8mm mm mm Insulation Lime Historical Clinker system Brick cement plaster Lime cement plaster In the graphs in Figure, the simulation results are represented in terms of relative humidity and water contents as function of location and time. The water content fields on the right side document the drying stage of the three build-ups reached at Dec.3 after years drying time. The image presentation allows a quick overview of the moisture situation and gives a descriptive impression about the differences in drying behavior of the constructions. For the reference build-up without insulation, moisture distribution has reached a maximum between outer clinker layer and inner brick layer. While drying can take place in both directions, the only possible moisture penetration is that from caused by rain. Obviously, the drying process dominates the water penetration by rain; otherwise no drying would have taken place. It must be concluded that the reasons for the high initial moisture contents found in Rijksmuseum must lay in imperfections of the envelope walls that could have forwarded a moisture penetration. The higher moisture level in the build-up with cellular glass insulation can be immediately recognized. The diffusion-tight barrier formed by the cellular glass prevents any drying towards the interior side. The drying potential towards exterior is limited due to the temperature drop in the wall cross section. This leads to an increase in moisture contents up to Vol% at the interface between existing construction and applied insulation layer. It is remarkable that this increase still exists after five years drying time. The build-up with diffusion-open insulation provided by calcium silicate shows a moisture response comparable to the reference construction without insulation. The maximum in moisture contents forms again between outer clinker layer and inner brick layer. This is of particular interest since it indicates that interstitial condensation (to be expected here between calcium silicate glue mortar) is very little in terms of moisture contents compared to the moisture penetration due to driving rain. A more detailed quantitative image give the graphs on the left side of Figure where the relative humidity profiles are depicted during a full course over the fifth year. The black range indicates the local variation of the relative humidity over the year at the respective position. Zones in which the relative humidity exceeds the 9%-limit at least once a year are considered temporarily wet. A maximum overhygroscopic penetration depth can be introduced indicating the width of the zone subjected to higher moisture contents. The heat losses in this range increase due to the higher thermal conductivity of wet materials and the resistance against frost damages will be lowered according to the rise of the freezing point temperature. The graphs support the conclusions drawn from the moisture fields. The reference buildup and the calcium silicate version behave comparably while the cellular glass variant stays wet. In addition, there is a remarkable improvement of the local climate at the internal side for both insulated variants, indicated by the diminished variation of the surface relative humidity. At the interface between insulation and glue mortar, calcium silicate shows the highest variation.

Relative humidity profiles Water content fields. Reference case without insulation LC-Plaster Brick LC-Plaster Clinker Existing construction, Water content in Vol% 9 9 8 8 7 7 6 6 Maximum overhygroscopic penetration depth 3 Location in mm 6 3 6. 3. 3... 9. 9 8. 8 7. 7 6. 6.. 3. 3.... Build-up with 3mm cellular glass insulation Cellular glass insulation + glue mortar Clay plaster LC-Plaster Brick LC-Plaster Clinker 9 9 8 8 7 7 6 6 Maximum overhygroscopic penetration depth 3 Location in mm 6 Cellular glass, Water content in Vol% 3 6. 3. 3... 9. 9 8. 8 7. 7 6. 6.. 3. 3... 3. Build-up with 3mm calcium silicate insulation Calcium silicate insulation + glue mortar LC-Plaster LC-Plaster Brick LC-Plaster Clinker 9 9 8 8 7 7 6 6 Maximum overhygroscopic penetration depth 3 Location in mm 6 Calcium silicate, Water content in Vol% 3 6. 3. 3... 9. 9 8. 8 7. 7 6. 6.. 3. 3... Figure. Profiles of relative humidity versus location during the th year (left) and water content fields at the end of five years simulation (right). Existing construction (top), 3mm foam glass insulation (middle) and 3mm calcium silicate insulation (below).

The integral moisture mass of the wall cross section is an indicator for the long term behavior. The comparison of the three build-ups, shown in Figure 6, demonstrates a clear difference between cellular glass and calcium silicate. While calcium silicate follows closely the drying curve of the reference case without insulation, the cellular curve shows a distinct course. The application of cellular glass decreases the drying potential of the construction and retards the drying process remarkably. After five years, the integral moisture content remains three times higher than in the constructions with maintained drying potential. After five years the build-ups have nearly reached a quasi-stationary state in which the annual average rain penetration approximately equals the drying rate. That means the constructions will retain their moisture levels reached in Figure 6. This level is inacceptable high for the cellular case and the probability for consecutive moisture damages raises. Integral water mass in kg/m 9 8 7 6 3 Without insulation 3mm Cellular glass insulation 3mm Calcium silicate insulation 3 Figure 6. Integral moisture content for different build-ups over years Reminding the improvement of the internal micro climate near the wall surface provided by both insulation variants and the long term drying behavior, where cellular glass fails, the comparison of the results justifies the conclusion that calcium silicate insulation would be the most appropriate solution. But the thermal insulation value of calcium silicate is not as high as the one of cellular glass (compare thermal conductivity values in Table ). Therefore the consequences of an increased thickness of the insulation layer were investigated. Integral water mass in kg/m 8 7 6 3 9. 3mm Calcium silicate insulation mm Calcium silicate insulation mm Calcium silicate insulation..6 Figure 7. Integral moisture content for different calcium silicate thicknesses in the th year.8 The curves in Figure 7 show the integral moisture contents of build-ups with 3mm, mm and mm thick calcium silicate insulation. While the curves for 3mm and mm are practically identical, a slight increase can be noticed for mm. This slightly increased moisture content is interstitial condensation that forms due to the lowered temperature in the condensation layer between calcium silicate and glue mortar. However, the course of the integral water mass is dominated by rain penetration.. Evaluation of the condensation risk During the heating period, especially on the of wall surfaces and on the cold side of the insulation layer, the occurrence of condensation becomes probable. Condensation can form as surface condensation or as interstitial condensation. For evaluation of the condensation risk, overhygroscopic water mass by condensation and by rain penetration must be distinguished. Therefore, condensate is defined here as a moisture content that is not caused by rain penetration and that exceeds the natural moisture content of the material in equilibrium to a relative humidity of 9%. The graphs in Figure 8 show the amount of interstitial condensate as function of time for different thickness of calcium silicate. The higher condensation values during the first two years are caused by redistribution of the initial moisture of the brick layer. After decaying of the high initial values, the condensate is constantly less than 3 g/m, which can be regarded as harmless. Condensate in kg/m...9.6.3 3mm Calcium silicate insulation mm Calcium silicate insulation mm Calcium silicate insulation 3 Figure 8. Amount of interstitial condensate for different thicknesses of the calcium silicate layer over years The risk of mould growth and surface condensation can be evaluated knowing the temperatures and relative humidity on the inner surface of the walls. Apart from a lower risk of mould growth and condensation, higher surface temperatures also create a more comfortable room climate and better conditions for art storage. A lower difference between room and surface temperature will also cause lower airflow velocity in rooms.

Figure 9 shows the course of the relative humidity on the inner wall surfaces during the fifth year. With none of the listed constructions surface condensation is predicted. The risk of mould growth is also low for the existing built up as well, because very few mould species are able to grow at these courses of relative humidity. But these conclusions do not refer to thinner walls and thermal bridges like room corners, where the temperature is lower and the relative humidity is higher than in undisturbed wall areas. Condensate in kg/m.8.6.. 3mm Calciumsilikat on mm brick wall 3mm Calciumsilikat on 6mm brick wall 3 8 8 7 7 6 6. 6mm brick wall without insulation 3mm Cellular glass on 6mm brick wall 3mm Calcium silicate on 6mm brick wall..6 Figure 9. Course of relative humidity on inner wall surface during the th year for a brick wall of 6mm thickness If a construction with mm brick is considered the need for internal insulation becomes more evident, see Figure. A difference of about % in relative humidity and 3.3 C in temperature between the current situation and the build-up with foam glass can be observed in this case. At the end of December, the surface of the existing built up cools down to C. Here, mould growth and surface condensation is likely, in particular at critical construction details. 8 8 7 7 6 6...6 A comparison of the interstitial condensate between the mm and 6mm brick walls is shown in Figure. The risk of condensation normally increases with lower wall thickness. Here, the mm brick wall shows slightly increased values but lays still in the harmless range of max.. kg/m. In the first winter, it falls below the curve of the 6mm wall due to its faster drying behavior..8 mm brick wall without insulation 3mm Cellular glass on mm brick wall 3mm Calcium silicate on mm brick wall Figure. Course of relative humidity on inner wall surface during the th year for a brick wall of mm thickness.8 Figure. Course of interstitial condensate during the years for brick walls of different thicknesses. Reduction of thermal energy transmission The heat transfer coefficient (U-value) characterizes the thermal transmission through the envelope walls. Under stationary conditions, the U-value is calculated from the thermal conductivities and the layer thickness of the materials taking into account the convective heat exchange coefficients at the surfaces. The transient U-value can be derived from an average heat flux over the inner wall surface divided by an average (internal-external) temperature difference. Both quantities are averaged over the heating period under consideration (here Oct.- to Mar-3). Table shows the transient U-values of tested wall constructions (temperature difference ΔT= 6.36 C) and the reduction of thermal energy transmission in percent. Table. Transient U-values of different build-ups based on 6mm brick wall Without insulation Cellular glass 3 mm Calcium silicate 3 mm Calcium silicate mm Calcium silicate mm Average heat flux (W/m ) U-value (W/m K) 8.9. -.8.66..76 33.9.67 9.96.6 6 Reduction of thermal energy transmission (%) All insulation variants reduce the transmission of thermal energy remarkably (up to 6%). From this point of view, a thermal insulation is recommended. It can be noticed that mm calcium silicate achieves the same reduction as 3mm cellular glass. But the advantage of mm calcium silicate over mm would be not very big. A further increase of the thickness would lower the cost-value ratio and can be regarded as inefficient.

6 SUMMARY AND CONCLUSIONS The building physical state of the Rijksmuseum Amsterdam caused serious concerns due to higher moisture contents in the envelope parts and unfavorable micro climatic conditions near the internal wall surfaces. The need for application of an internal thermal insulation has been commonly recognized ARUP/DGMR (). It was to decide which insulation option would be most appropriate to regulate the hygrothermal behavior of the exterior walls according to the requirements of the exhibits. Two insulation options cellular glass and calcium silicate were discussed and both were compared with the reference case original construction without insulation. The solution was achieved by using the method of numerical simulation that yields the temporal courses of temperature, relative humidity, water content and condensate. Statements about advantages and disadvantages of the insulation options can be derived from that.. Envelope walls of a thickness d 6mm should be generally insulated. If paintings are attached, the walls should be insulated up to a thickness of 9mm. This has positive effects on the micro climate near the internal surfaces. Application of an insulation layer of minimum 3mm thickness would keep the relative humidity permanently below 7%.. The application of foam glass forms a vapor barrier at the internal surfaces preventing drying towards the inner surface. The drying potential of wet building elements is drastically decreased. Thus, the moisture equilibrium of the envelope walls between rain penetration and evaporation adjusts at an up to three times higher moisture level. 3. With calcium silicate insulation the drying potential of the envelope walls is kept. Sufficiently high temperatures and low relative humidity on the interior surface would prevent mould growth and surface condensation. As a result indoor climate quality would also be improved.. The favorite build-up is an insulation system with mm calcium silicate. The thermal transmission through the walls can be reduced by about %. A higher insulation layer thickness would lower the cost-value ratio. The analysis of the undisturbed wall areas has been complemented by two-dimensional analysis of critical construction elements. Since the results shown here were conservatively interpreted they could be confirmed by the results of the D-simulations. In addition, the increased risk of frost damages at thin wall structures was assessed. The freezing point temperature in porous media is decreased in dependence on the moisture content, Xu (998). The increase of frost damage risk by calcium silicate insulation was found to be very low. The temperature drop in the outer parts of the wall cross section was lower than K. The simulations were based on material data from the simulation program s data base and from literature. To justify the results, the material parameters of bricks and mortar from the Rijksmuseum were determined in the Laboratory of the Institute of Building Climatology. Repeated simulations with updated material data confirmed the results and conclusions. Additionally, a mock-up area was built up in the Rijksmuseum where calcium silicate insulation was applied. Sensor technology for measurement of relative humidity, temperature and heat fluxes has been installed. Continuous monitoring of the hygrothermal behavior by in-situ measurements is in progress. 7 REFERENCES Arup DGMR, : Rijksmuseum Amsterdam - Preliminary Design (VO) Report Building Physics. Arup BV, Issue 3, Dec., Job Number 98 Arup DGMR, : Personal communication between Arup BV and the Authors: Build-ups -dimensional analysis and Climatic data from Amsterdam, received by e-mail. Grunewald, J. 997: Diffusiver und konvektiver Stoff- und Energietransport in kapillarporösen Baustoffen. Dissertation, Dresden University of Technology, Institute Building Climatology Grunewald J., R. Plagge, P. Häupl, : A -levelled Hygrothermal Material Database for the Numerical Simulation Program DELPHIN, 6th Symposium on Building Physics in the Nordic Countries, Trondheim, Norway Grunewald J., P. Häupl, 3: Gekoppelter Feuchte-, Luft-, Salz-, und Wärmetransport in porösen Baustoffen. Bauphysik-Kalender 3. Berlin: Ernst & Sohn Verlag Scheffler, G., Grunewald, J : Calibration of an Engineering Model of hygrothermal material characterisation, CIB W Meeting in Glasgow Sedlbauer, K.. Vorhersage von Schimmelpilzbildung auf und in Bauteilen. Dissertation, Universität Stuttgart Seifert, B. et al. : Leitfaden zur Vorbeugung, Untersuchung, Bewertung und Sanierung von Schimmelpilzwachstum in Innenräumen ( Schimmelpilz-Leitfaden ). Berlin: Umweltbundesamt (UBA) Berlin Xu, Y. 998: Numerische Simulation der Eisbildung in kapillarporösen Baustoffen unter Berücksichtigung der gekoppelten Wärme- und Feuchtetransportprozesse. Dissertation, in: Dresdner Bauklimatische Hefte, Heft, TU Dresden, Dresden