Control of Winter Surface Condensation in a Tissue Paper Production Hall: A Measurement-Calibrated, TMY-Based Comparative Assessment

1. Introduction

Surface condensation inside industrial buildings is the visible outcome of a simple inequality — the internal surface temperature of the envelope falls below the dew point of the adjacent air — but its consequences are neither simple nor cheap. Persistent liquid water on walls and ceilings corrodes reinforcement and fixings, saturates finishes, drips onto product and equipment, and sustains microbial colonisation of the surfaces it wets. Aelenei and Henriques [1] showed that condensation risk analysis on building envelopes must be treated as a coupled thermal and hygric problem rather than as a check on a single design condition, because the risk is governed by the joint distribution of surface temperature and vapour content over the whole season. Xue et al. [2] demonstrated experimentally that moisture accumulation in envelopes is self-reinforcing: once a surface has been wetted, its capacity to release the accumulated moisture is reduced, so condensation and mould growth become progressively easier to trigger. Orlik-Kożdoń [3] quantified the sensitivity of both water-vapour condensation and mould germination to indoor climatic conditions, confirming that the indoor humidity level, not the indoor temperature alone, is the controlling variable. The review of moisture ingress in building envelope materials by Lopez-Carreon et al. [4] places surface condensation within the broader family of moisture transport mechanisms and stresses that mitigation must act either on the vapour supply or on the surface temperature, the material itself offering little defence once wetting is established.

The design rules that follow from this understanding are codified in EN ISO 13788 [5], which defines the internal surface temperature required to avoid critical surface humidity through the temperature factor f_Rsi, and in EN ISO 6946 [6], which fixes the surface resistances used to compute it. Ilomets and Kalamees [7] applied the same criterion to thermal bridges and showed that the criticality of a junction is decided by the ratio between the internal surface resistance and the total resistance of the assembly: where that ratio is large — that is, where the construction itself is thermally weak — the surface temperature is effectively slaved to the outdoor temperature and no realistic increase of indoor temperature can lift it above the dew point. This mechanism, well documented for thermal bridges in insulated buildings, applies a fortiori to entire uninsulated elements such as the bare reinforced-concrete walls typical of 1970s Eastern European industrial construction. Oladokun et al. [8] coupled computational fluid dynamics with a mould growth model and confirmed that the local surface temperature field, rather than the bulk room condition, determines where colonisation begins.

Large industrial and sports halls introduce a further complication: they are strongly stratified. Karpuk et al. [9] measured vertical air temperature gradients in a large industrial hall and found differences of the order of 10 K between the occupied zone and the roof zone, so that a single well-mixed node is a poor representation of the enclosure. Lin et al. [10] extended this observation to humidity, showing for ice arenas that thermal and hygric stratification are statistically distinct and that the upper zone can be simultaneously the warmest and the most vapour-laden part of the volume — precisely the combination that maximises condensation risk on the roof and on the upper part of the walls. Limane et al. [11] reached comparable conclusions for indoor swimming pools by CFD, and the systematic review of swimming facilities by Smedegård et al. [12] identified envelope condensation control as the dominant constraint on the ventilation and dehumidification strategy of high-latent-load enclosures. Swimming halls are, in this respect, the closest well-studied analogue of a tissue mill: a large volume with an intense, continuous and spatially concentrated internal moisture source.

In the pulp and paper sector the moisture source is the drying section. Hubbe [13] reviewed the energy efficiency challenges of pulp and paper manufacturing and identified drying as the single largest consumer, while Laurijssen et al. [14] benchmarked energy use at process-unit level across European mills and reported wide dispersion in the performance of the drying and hood-exhaust systems. Tissue machines are the most extreme case, combining a Yankee cylinder with a high-temperature hood; Yan and Li [15] simulated heat-pump recovery from paper exhaust hoods and confirmed that the exhaust stream carries a large, low-grade thermal and latent load. When the hood capture efficiency falls short — through leakage, fouling or under-sized extraction — the uncaptured fraction is released directly into the hall and becomes an internal moisture gain of several tonnes per day. Despite the industrial importance of the problem, published quantitative assessments of condensation control in paper mill halls are scarce, and the hood spillage that drives it is seldom reconciled against an independent measurement of the hall air moisture balance.

Where the moisture cannot be removed at source, the vapour content of the air must be lowered. Ma et al. [16] formulated the choice of supply-air humidity as an optimisation over spatially distributed moisture demands and showed that, where the requirement is set by the most critical location rather than by the mean condition, the design target must be derived from that location — for a condensation-prone hall, from the coldest surface. Mechanical dehumidification by cooling coil is limited by the apparatus dew point: the coil cannot dry air below the saturation humidity ratio corresponding to its surface temperature, whatever the airflow. Solid desiccant wheels remove this ceiling by adsorbing vapour directly on a hygroscopic matrix. Xu and Chen [17] reviewed low dew-point desiccant wheel systems and mapped the trade-off between attainable dew point, regeneration temperature and specific energy; Chen et al. [18] demonstrated cascaded wheels delivering supply air below 0 °C dew point; Guan et al. [19] and Ma et al. [20] reported on-site performance of deep-dehumidification wheel installations in lithium-battery plants, where regeneration heat dominates the energy bill. Hybrid arrangements, in which a cooling coil performs the bulk moisture removal and a wheel finishes the drying, were analysed by Jia et al. [21] and, for buildings with high latent loads, by Comino et al. [22], who reported substantial reductions of regeneration energy relative to a wheel operating alone. These studies establish the technical feasibility of deep dehumidification but also its cost: regeneration heat is the most expensive energy in the chain, and its magnitude scales with the moisture actually passed to the wheel.

Finally, a credible comparison between solutions that behave differently across the weather spectrum cannot rest on a single design point. Sun et al. [23] discussed the generation and use of typical meteorological years for coupled building simulation, and hourly TMY data are now routinely available from the PVGIS service of the Joint Research Centre [24] for any European location. Hour-by-hour evaluation over a full year is the only way to convert a physical criterion into an operational metric — hours of condensation per season — and to weight energy consumption by the frequency of the conditions that cause it. Freimanis et al. [25] assessed seventeen insulation systems applied to solid masonry walls and found that the hygrothermal risk transferred to the wall depends strongly on where the insulation is placed, external systems being the safest option for an enclosure whose inner face must remain dry. On the economic side, Basińska et al. [26] showed for building retrofit that insulation measures are frequently justified on energy grounds alone, independently of any secondary benefit.

The present work brings these strands together for a specific and, in Eastern Europe, widespread configuration: a 1975 reinforced-concrete tissue mill hall with an uninsulated envelope, a heavily loaded ventilation system and chronic winter condensation on the upper walls and the false ceiling. The objectives are (i) to close the hall moisture balance against measurements and quantify the discrepancy with the declared process water balance, (ii) to identify the physical mechanism responsible for winter condensation, (iii) to evaluate three solution families — envelope insulation, ventilation intensification and mechanical dehumidification — on a single hourly TMY basis using identical technical, energetic and economic criteria, and (iv) to establish whether these solutions are alternatives or whether one of them dominates.

2. Material

2.1. The Building and the Process

The facility is a thin-paper (tissue) mill located in Petrești, Alba County, Romania (45.921° N, 23.561° E, 278 m a.s.l.), commissioned in 1975. The structure is reinforced concrete, partly precast, with a 30 m central span housing the paper machines and two 6 m lateral spans on four levels. The central span is single-storey plus a technological floor at +6 m; the reinforced-concrete slab comprises a 280 mm plate, 300 × 800 mm secondary ribs and 500 × 1200 mm main beams at 6 m centres. The analysis boundary is the production hall itself: a main volume of 80 × 30 × 14.5 m (2 400 m², 34 800 m³) and an adjoining auxiliary volume of 23 × 30 × 6 m (690 m², 4 140 m³), giving a total enclosed volume of 38 940 m³ and a total external wall area of 3 190 m² (220 m perimeter × 14.5 m height).

The envelope is a 200 mm reinforced-concrete wall without insulation. With λ = 2.3 W m⁻¹ K⁻¹, R_si = 0.13 m² K W⁻¹ and R_se = 0.04 m² K W⁻¹ [6], the total thermal resistance is 0.257 m² K W⁻¹ and the thermal transmittance 3.89 W m⁻² K⁻¹. The intrinsic resistance of the concrete layer alone is 0.087 m² K W⁻¹ — one third of the total — so the assembly is, thermally, little more than a pair of surface films. A false ceiling separates the hall from a ventilated plenum beneath the roof.

Production is carried out on two tissue machines, denoted PM6 and PM7 in the ventilation documentation. The process is water-intensive; part of the process water reaches the hall air as vapour. The process water balance supplied by the operator is summarised in Table 1. It attributes to the hall air only 4–5 t d⁻¹ from PM6 and 1–2 t d⁻¹ from PM7, the bulk of the evaporated water being extracted through the machine hoods (65–70 and 100–105 t d⁻¹ respectively) and, for PM7, through a dedicated mist extraction system (10–15 t d⁻¹).

2.2. The Ventilation System

Fresh air is supplied along the long sides of the hall at machine level and, separately, into the false-ceiling plenum; air is extracted from the short side of the hall immediately below the false ceiling. Each machine is served by its own supply system. The PM6 supply air handling unit is fitted with a steam heating coil which is currently inoperative: the steam supply and condensate lines have been cut and the coil is not connected to any heat source. The PM7 supply system incorporates two recuperative heat exchangers that preheat the fresh air with the hot exhaust from that machine — one for the air delivered to the hall, one for the air delivered to the plenum.

The hall supply fans are centrifugal units of type CHB 49, the plenum supply fan is a CHB 36. From the manufacturer's characteristic curves, and allowing for the elevated system resistance caused by paper-dust deposits on intake filters and ductwork, the fans are estimated to operate at 600–750 rpm and to deliver 40 000–45 000 m³ h⁻¹ each, i.e. 80 000–90 000 m³ h⁻¹ in total. Extraction is provided by two fans per machine (⌀ 1 200 mm) plus three lateral fans (⌀ 1 000 mm). The plenum supply fan was observed running at 300 rpm, below the lowest speed shown on its characteristic diagram, corresponding to an estimated 10 000–15 000 m³ h⁻¹.

A cross-section of the ventilation arrangement is shown in Figure 1.

2.3. Efficiency Measures Considered

Five measures were placed on the table at the outset.

Source reduction. Improving hood capture and machine exhaust is the only measure that acts on the cause rather than on its effects, and it is therefore the first priority. Its investigation requires process expertise specific to the machine suppliers and falls outside the scope of the present study; it is recommended to the operator as a parallel action.

External envelope insulation. Adding expanded polystyrene (EPS, λ = 0.038 W m⁻¹ K⁻¹) to the outer face of the concrete wall raises the internal surface temperature without altering the indoor moisture balance.

Ventilation intensification. Winter outdoor air carries 1–3 g kg⁻¹ of moisture, i.e. it is drier than the air a cooling coil can produce. Increasing the airflow dilutes and exports the process moisture and lowers the indoor dew point without any dehumidification equipment, at the cost of heating the additional fresh air.

Mechanical dehumidification. Three configurations were considered for treating the upper-zone air: a cooling coil alone, a solid desiccant wheel alone, and a hybrid arrangement in which the coil performs the bulk drying and the wheel finishes it (Figure 2). The coil can be fed either by a dry cooler in free-cooling mode or by a chiller; the wheel is regenerated with air heated to 120–150 °C, potentially recovered from the machine exhaust.

False-ceiling heating. The temperature of the false ceiling follows the temperature of the plenum air above it. At present only PM7 contributes recovered heat to the plenum. Extending the same recuperative arrangement to PM6 would widen the achievable range of ceiling temperature. For an estimated plenum airflow of 10 000–15 000 m³ h⁻¹ (≈ 15 000 kg h⁻¹ = 4.16 kg s⁻¹) heated from −15 °C to 35 °C, the required thermal power is ≈ 230 kW. Because the source is the hot exhaust of PM6, the measure does not increase the site energy consumption.

3. Method

3.1. Psychrometric Relations

The saturation vapour pressure was computed with the improved Magnus form of Alduchov and Eskridge [27]:

p_sat(t) = 610.94 · exp [17.625 t / (t + 243.04)] [Pa] (1)

with the partial vapour pressure p_v = (φ/100) p_sat, the humidity ratio

x = 0.622 p_v / (p − p_v) [kg kg⁻¹] (2)

and the dew-point temperature obtained by inversion of (1),

t_d = 243.04 α / (17.625 − α), α = ln(p_v / 610.94) [°C] (3)

The moist-air density and specific enthalpy were evaluated as

ρ = [p / (R_a T)] · (1 + x) / (1 + 1.608 x) [kg m⁻³] (4)

h = c_pa t + x (r_v + c_pv t) [kJ kg⁻¹ of dry air] (5)

with Ra = 287.055 J kg⁻¹ K⁻¹, cpa = 1.006 kJ kg⁻¹ K⁻¹, cpv = 1.863 kJ kg⁻¹ K⁻¹ and rv = 2 500 kJ kg⁻¹. A barometric pressure of 101 325 Pa was used throughout.

3.2. Moisture Balance of the Hall

The water taken up by the hall air follows from the exhaust airflow and the increase of the humidity ratio between supply and exhaust:

ṁ_a = V̇ ρ / (1 + x_exh), ṁ_w = ṁ_a · Δx, Δx = x_exh − x_sup [kg h⁻¹] (6)

Exhaust airflows were obtained from velocity traverses at each fan, using the measured duct diameters. Because the process releases the same quantity of water in summer and in winter, Δx was treated as a property of the process and of the installed airflow, and was carried over unchanged to the winter analysis. The same reasoning was applied to the sensible temperature rise of the air across the hall, ΔT.

3.3. Surface Condensation Criterion

For steady one-dimensional heat flow, the total thermal resistance and the internal surface temperature of an opaque element are

R_tot = R_si + Σ d_i/λ_i + R_se, U = 1/R_tot (7)

t_si = t_i − (R_si / R_tot)(t_i − t_e) (8)

Condensation is predicted whenever t_si < t_d. Equation (8) can be rewritten as a weighted mean,

t_si = (1 − f) t_i + f t_e, f = R_si / R_tot (9)

in which f is the fraction of the total temperature drop consumed by the internal surface film. For the existing wall f = 0.13/0.257 = 0.506; adding 50 mm of EPS raises R_tot to 1.573 m² K W⁻¹ (U = 0.636 W m⁻² K⁻¹, a reduction of 84 %) and lowers f to 0.083.

The EPS thickness required to avoid condensation with a safety margin ΔT follows from the condition t_si ≥ t_d + ΔT:

R_tot,nec = R_si (t_i − t_e) / (t_i − t_d − ΔT) (10)

d_EPS = (R_tot,nec − R_si − d_c/λ_c − R_se) · λ_EPS (11)

A margin ΔT = 1 K was adopted. Equation (10) becomes singular as t_d approaches t_i: when the indoor dew point exceeds the indoor air temperature, no thickness of insulation can prevent condensation, because the surface temperature can never exceed the air temperature that drives it.

3.4. Fog Formation Criterion

Fog inside the hall was checked separately. Mixing a fraction f_c of cold outdoor air, infiltrated through doors and leaks or introduced unheated, with warm humid indoor air gives

t_mix = f_c t_e + (1 − f_c) t_i, x_mix = f_c x_e + (1 − f_c) x_i (12)

and fog appears if x_mix > x_sat(t_mix), which is possible even when both components are unsaturated because the saturation curve is convex.

3.5. Zone Model and Parametric Analysis

The hall was represented by two zones — a lower zone at machine level and an upper zone beneath the false ceiling — because vertical stratification of both temperature and humidity is well established for enclosures of this size [9,10]. The process moisture is released in the lower zone, rises with the thermal plumes of the machines and accumulates in the upper zone, where the exhaust openings are located; the upper-zone humidity ratio was therefore taken equal to the measured exhaust value. A winter reference case was defined as: process moisture release 27.2 t d⁻¹ (measured), outdoor air −5 °C and 85 % RH, lower zone 25 °C, upper zone 35 °C, false-ceiling surface 20 °C. Each influencing parameter was then varied in turn with the others held at their reference values.

3.6. Hourly Seasonal Analysis

The parametric analysis was extended to a full year using a typical meteorological year generated by PVGIS from the 2005–2023 record for the site coordinates [24], giving 8 760 hourly values of dry-bulb temperature, relative humidity and pressure. The cold season was defined by t_e < 12 °C, which yields 5 053 hours per year (58 % of the year); above this threshold no condensation is predicted, in agreement with plant observation. For each hour the outdoor humidity ratio was computed from (2), the indoor humidity ratio as x_i = x_e + Δx, the indoor dew point from (3), the surface temperature from (8), and the hour was flagged as a condensation hour if t_si < t_d. The metrics reported are the number of condensation hours per season and the minimum margin t_si − t_d over the season.

For the ventilation option, the airflow required in each hour is that which reduces Δx sufficiently to bring t_d below t_si with the 1 K margin; the associated energy is the sensible heat needed to raise the additional outdoor air to the hall temperature. For the dehumidification options, the duty is not the total process generation of 27.2 t d⁻¹ but the residue that the existing ventilation fails to export. The cooling coil was characterised by its apparatus dew point (ADP ≈ 9 °C), which imposes a drying floor x_ADP = x_sat(9 °C) = 7.4 g kg⁻¹; the desiccant wheel by its regeneration power, computed from the mass of water it must adsorb; and the hybrid by the split of the total moisture load between the two components. Free-cooling availability was evaluated from the dry-cooler approach: with an 8 K approach the required 6.5 °C glycol supply is attainable only below t_e ≈ −1.5 °C.

3.7. Energy and Economic Evaluation

Seasonal transmission losses through the wall were computed as U·A·Σ(t_i − t_e) over the 5 053 cold-season hours with A = 3 190 m². The insulation option was charged with the additional heating required to raise the hall from 25 to 27 °C and credited with the reduction of transmission losses. Energy was valued at 60 € MWh⁻¹ (thermal) and 150 € MWh⁻¹ (electrical). Investment costs were estimated at feasibility-study level with an accuracy of ±40–50 %, and a ten-year total cost of ownership was computed without discounting. The whole wall area was insulated rather than only the upper zone affected by condensation, because the heat saving is proportional to area and the payback period is therefore identical for the lower portion, which consequently finances itself.

4. Results and Discussion

4.1. Airflow and Moisture Balance

Two measurement campaigns were carried out, the first on 17 June with only PM6 running, the second on 2 July with both machines in operation. Velocity traverses at the extraction fans gave a mean of 10 m s⁻¹ at the four machine fans (65 144 m³ h⁻¹) and 6 m s⁻¹ at the three lateral fans (20 358 m³ h⁻¹), a total exhaust airflow of 85 502 m³ h⁻¹. This is consistent with the 80 000–90 000 m³ h⁻¹ estimated independently from the supply-fan characteristic curves, and the extraction measurement was therefore not repeated in the second campaign. The corresponding air change rate is 2.4–2.5 h⁻¹.

The measured air states are given in Table 2. The consistency between the humidity ratio measured at the exhaust and that measured at various points inside the hall (17.0 g kg⁻¹ against 18.0 g kg⁻¹ in the first campaign, 21.5 g kg⁻¹ against 26.4 g kg⁻¹ in the second) confirms that the exhaust value is a defensible proxy for the upper-zone condition; the in-hall points were affected by adjacent spaces (a partly open door at the walkway, the stairwell at the balcony) and were used only as a plausibility check.

The resulting moisture balances are given in Table 3. With one machine running the hall air takes up 18.4 t d⁻¹; with both machines, 27.2 t d⁻¹. These figures exceed the declared release into the hall (Table 1: 5 and 7 t d⁻¹ respectively) by factors of 3.7 and 3.9. The near-constancy of the ratio across two different operating states argues against random measurement error and points to a systematic discrepancy, attributable to some combination of an under-estimated declared release, malfunctioning hood extraction, poor local capture of the vapour plume, and leakage returning extracted vapour to the hall. Interpreted as a capture deficiency, the measurements imply that about one tenth of the vapour nominally extracted by the hoods and the mist system (≈ 190 t d⁻¹ for the two machines) ends up in the hall air instead. This finding alone justifies the priority given to source reduction, and it is also the reason why the condensation problem is so severe: the same building with a properly captured process would not condense.

4.2. Winter Sensitivity Analysis

With Δx = 12.2 g kg⁻¹ carried over to winter conditions, the reference case (t_e = −5 °C, φ_e = 85 %) gives an upper-zone humidity ratio of 14.4 g kg⁻¹ and an upper-zone dew point of 19.7 °C. At the reference false-ceiling temperature of 20 °C condensation on the ceiling is avoided only marginally, and appears as soon as the ceiling falls below 19.7 °C; on the upper part of the walls condensation occurs.

The parametric sweeps identified a clear hierarchy of influence. The process moisture release is by far the strongest driver: it acts directly on the upper-zone humidity ratio and hence on the dew point, and higher releases demand correspondingly higher surface temperatures. The outdoor air temperature acts on both zones, since the moisture released by the process is added to that already carried by the supply air; a warmer outdoor air at constant relative humidity carries more absolute moisture and raises the indoor dew point. The outdoor relative humidity has only a weak effect, because in winter the absolute moisture content of outdoor air is small at any relative humidity. The indoor air temperatures — lower zone and upper zone — have no effect at all on the dew point, since they do not change the humidity ratio; they act only through the surface temperature, and raising the upper-zone air temperature over the range investigated was found insufficient to lift the wall out of the condensation regime. The false-ceiling temperature, finally, does not affect the dew point but is decisive for condensation on the ceiling itself. In the lower zone no condensation risk was found under any of the conditions investigated.

4.3. The Mechanism of Winter Condensation

The reason the uninsulated wall cannot be rescued by heating follows directly from equation (9). With f = 0.506, the surface temperature is very nearly the arithmetic mean of the indoor and outdoor air temperatures:

t_si = 0.494 t_i + 0.506 t_e

In summer this is harmless. At t_e = 30 °C with the hall at 38.5 °C the wall surface reaches 34.2 °C while the indoor dew point is 29.5 °C — a margin of 4.7 K, and no condensation, exactly as observed in the plant. In winter the same equation is fatal. At t_e = −5 °C the indoor dew point is 19.7 °C while the surface temperature of a hall held at 30 °C falls to 12.3 °C. To lift the surface to the dew point the hall would have to be heated to about 45 °C, and to about 52 °C at t_e = −15 °C. Winter condensation on this wall is therefore unavoidable by heating alone.

It is worth stating the corollary explicitly: the phenomenon requires both factors. The near-zero thermal resistance of the wall is one; the elevated dew point produced by the process moisture is the other. In the absence of the process release the indoor dew point at the reference outdoor condition would be about −7 °C, and the bare concrete wall would not condense at all.

Figure 3 shows the surface temperature, the indoor dew point and the indoor air temperature as functions of outdoor temperature for the uninsulated wall and for the wall with 50 mm of EPS, at indoor temperatures of 25 and 35 °C. On the uninsulated wall (panels a and b) the surface line lies below the dew-point line over the whole cold range at both indoor temperatures. With 50 mm of EPS (panels c and d) the surface line is lifted above the dew point; at 25 °C, however, the two lines cross at about 12 °C outdoor temperature, beyond which the indoor dew point exceeds the indoor air temperature itself and condensation resumes regardless of insulation thickness — the singularity of equation (10). At 35 °C the margin is comfortable across the whole range.

Figure 4 gives the EPS thickness required by equations (10) and (11). Except in the region where insulation cannot help at all, the required thickness stays below 50 mm; 50 mm of EPS was therefore adopted as the design value.

4.4. Solution 1 — Envelope Insulation

Table 4 gives the hourly evaluation over the 5 053 cold-season hours. The result that matters is not that insulation works, but that insulation and heating are complementary rather than alternative. Insulation with a cold hall (20 °C) leaves 5 051 condensation hours; heating alone on the bare wall leaves 4 605 hours even at 35 °C. The combination removes condensation entirely, with a sharp threshold at 50 mm and 27 °C, where the minimum seasonal margin turns positive (+0.6 K).

The same result is shown in Figure 5. The residual 239 hours at 25 °C correspond to the mildest part of the cold season, when the outdoor air is warm enough to carry appreciable absolute moisture into the hall; they disappear with a 2 K increase of the hall temperature.

4.5. Solution 2 — Ventilation Intensification

Winter outdoor air, at 1–3 g kg⁻¹, is drier than air leaving a cooling coil at its apparatus dew point, so increasing the airflow is itself a dehumidification strategy. Table 5 gives the airflow required to suppress condensation and the associated seasonal heating energy.

On the bare wall the measure amounts to doubling the ventilation plant — an additional 115 000 m³ h⁻¹ of air handling, ductwork and heating capacity — and to a seasonal energy surplus of about 4 000 MWh for heating the extra fresh air. On the insulated wall the required airflow falls below the airflow already installed, which shows that the existing plant, sized for a bare envelope, becomes over-sized once the envelope is treated.

4.6. Solution 3 — Mechanical Dehumidification

Table 6 gives the three configurations evaluated on the bare wall at unchanged ventilation airflow.

The cooling coil used alone is excluded on physical grounds. It cannot dry air below the saturation humidity ratio of its own surface, x_ADP = 7.4 g kg⁻¹ for an apparatus dew point of 9 °C, whatever the airflow. On the bare wall the target imposed by the cold surface falls below this floor for 139 hours per year with the hall at 30 °C, and in those hours — the coldest and most critical of the season — the coil cannot meet the requirement.

In series with a desiccant wheel the coil recovers its usefulness. It removes 45–51 % of the water in the range where it is efficient, from the hall humidity down to its own floor, and the wheel finishes the drying below that floor. Because the water taken by the coil requires no regeneration heat — the most expensive energy in the chain — the total consumption falls by 39–45 % relative to the wheel alone, in line with the reductions reported for hybrid systems in the literature [21,22].

Free-cooling contributes less than might be expected. Producing 6.5 °C glycol with a dry cooler at an 8 K approach requires an outdoor temperature below −1.5 °C, a condition met in only 681 of the 5 053 cold-season hours (13 %). Reducing the approach to 6.5 K extends free-cooling operation up to an outdoor temperature of 0 °C. The contribution is real but limited, and the principal merit of the hybrid arrangement remains the reduction of the regeneration duty rather than the elimination of the compressor.

The decisive observation, however, concerns the insulated wall: with 50 mm of EPS in place, the dehumidification duty is zero at every hall temperature analysed. All three configurations become superfluous.

4.7. Energy Balance

Table 7 gives the seasonal energy balance of the insulation option, charged with the heating needed to raise the hall from 25 to 27 °C and credited with the reduction of transmission losses over the full 3 190 m² of wall.

Insulation does not consume energy: it saves 826 MWh per season even after paying for the additional heating it requires. Transmission losses fall by a factor greater than five. Table 8 places this alongside the net consumption of the other options.

For solutions 2 and 3 the wall remains uninsulated, so the 1 306 MWh of transmission losses persist and are not included in the table; these options therefore also forgo the heat saving.

4.8. Economic Assessment

Investment estimates are of feasibility-study accuracy (±40–50 %) and do not replace tendering. Table 9 gives the capital cost of each option and the ten-year total cost of ownership at 60 € MWh⁻¹ thermal and 150 € MWh⁻¹ electrical.

Insulation is the only option with a negative total cost. The heat saving it produces, about 50 k€ per year, amortises the investment in 1.6–2.6 years depending on the unit execution price, and over ten years yields a net gain of about 394 k€, against costs of 3.6–4.1 M€ for the alternatives (Figure 6). The gap is an order of magnitude and is far larger than the ±40–50 % uncertainty of the estimates, so the ranking is robust.

It should be stressed that the wall insulation is economically justified as a pure energy-efficiency measure, independently of the condensation problem — a conclusion consistent with the retrofit economics reported by Basińska et al. [26]. That it simultaneously and completely resolves the condensation problem leaves it without a rational alternative.

4.9. Discussion and Limitations

Three points deserve emphasis. First, the two causes of the problem are separable and act multiplicatively: an envelope with negligible thermal resistance, and a dew point elevated far above the outdoor value by uncaptured process moisture. Acting on either one is sufficient in principle; acting on the envelope is by far the cheaper of the two available to the building designer, while acting on the moisture source is the responsibility of the process engineer and should be pursued in parallel. Second, the solutions are not independent alternatives to be traded off: insulating the envelope reduces the ventilation requirement below the installed capacity and reduces the dehumidification duty to zero. Third, the widely held expectation that a large industrial hall can be kept condensation-free by heating it harder is quantitatively wrong for an uninsulated heavy wall, and equation (9) shows why.

The analysis is subject to several limitations. The winter indoor air temperature was not measured; it was treated parametrically over 20–35 °C, and the condensation threshold emerged at 27 °C. Confirming the actual winter indoor temperature by measurement would refine the figures without altering the ranking of the options. The moisture increment Δx and the sensible temperature rise ΔT were calibrated on summer measurements and assumed constant into winter, on the argument that the process releases the same quantity of water year-round at unchanged airflow; a winter measurement campaign would test this assumption directly. The envelope model is steady-state and one-dimensional, and neglects thermal bridges, thermal mass and the moisture buffering of the concrete — all of which would make the real behaviour somewhat worse rather than better at thermal bridges and somewhat better in the short term through buffering. The two-zone representation is coarse compared with the CFD treatments applied to swimming halls and ice arenas [10,11], and does not resolve the local surface temperature field that governs where condensation first appears. Finally, the cost figures are order-of-magnitude estimates requiring confirmation by tender.

5. Conclusion

A tissue paper production hall suffering chronic winter condensation on its upper walls and false ceiling was analysed on the basis of two measurement campaigns and an hourly typical meteorological year, and three families of remedial solutions were compared on identical technical, energetic and economic criteria.

The measured exhaust airflow of 85 502 m³ h⁻¹ (2.46 h⁻¹) and the measured humidity-ratio increment of 12.2 g kg⁻¹ with both machines running correspond to a moisture uptake of 27.2 t d⁻¹ by the hall air — 3.9 times the value declared in the process water balance. About one tenth of the vapour nominally extracted by the machine hoods therefore reaches the hall air, and this uncaptured moisture is what raises the indoor dew point from about −7 °C to about 19.7 °C at the winter reference condition.

Condensation results from the conjunction of this elevated dew point with an envelope of negligible thermal resistance. Because the internal surface film carries 50.6 % of the total thermal resistance of the 200 mm reinforced-concrete wall, the surface temperature is approximately the mean of the indoor and outdoor temperatures. Additional heating alone is consequently not a solution: the bare wall would require an indoor temperature of about 45 °C at −5 °C outdoor and about 52 °C at −15 °C.

All three solution families can eliminate condensation completely, but they differ radically in investment and consumption. The cooling coil used alone is physically excluded, since the target imposed by the cold surface falls below its drying floor of 7.4 g kg⁻¹; in a hybrid arrangement with a desiccant wheel it takes about half the moisture load and reduces consumption by 39–45 % relative to the wheel alone. Ventilation intensification works without any dehumidification equipment, winter outdoor air at 1–3 g kg⁻¹ being itself a drying agent, but requires doubling the installed airflow and about 4 000 MWh of additional heat per season.

The recommended solution is external insulation of the walls with 50 mm of expanded polystyrene combined with maintaining the hall at a minimum of 27 °C. This eliminates condensation in all 5 053 cold-season hours, saves 826 MWh per season (about 50 k€ per year), amortises in 1.6–2.6 years and returns a net gain of about 394 k€ over ten years, against 3.6–4.1 M€ for the alternatives. Once the wall is insulated the dehumidification duty falls to zero and the required ventilation airflow drops below the installed capacity. Insulation does not compete with the other solutions; it makes them unnecessary.

Declaration on AI Usage: Artificial Intelligence (AI) tools were utilized to assist with text phrasing and refinement for linguistic accuracy, as well as to support the economic analysis presented in this study. All final outputs, interpretations, and conclusions remain the sole responsibility of the authors.