Air contaminants in a submarine equipped with air independent propulsion

Abstract

The Swedish Navy has operated submarines equipped with air independent propulsion for two decades. This type of submarine can stay submerged for periods far longer than other non-nuclear submarines are capable of. The air quality during longer periods of submersion has so far not been thoroughly investigated. This study presents results for a number of air quality parameters obtained during more than one week of continuous submerged operation. The measured parameters are pressure, temperature, relative humidity, oxygen, carbon dioxide, hydrogen, formaldehyde and other volatile organic compounds, ozone, nitrogen dioxide, particulate matter and microbiological contaminants. The measurements of airborne particles demonstrate that air pollutants typically occur at a low baseline level due to high air exchange rates and efficient air-cleaning devices. However, short-lived peaks with comparatively high concentrations occur, several of the sources for these have been identified. The concentrations of the pollutants measured in this study do not indicate a build-up of hazardous compounds during eight days of submersion. It is reasonable to assume that a substantial build-up of the investigated contaminants is not likely if the submersion period is prolonged several times, which is the case for modern submarines equipped with air independent propulsion.

---

Introduction

For more than two decades, the Swedish Navy has operated submarines with air independent propulsion (AIP) in the form of Stirling engines. Development of AIP submarines has increased the possible length of submersion from a few days to several weeks. The increase in submersion time puts new demands on the air management system on-board. Contaminants that were previously thought to be present only in minor concentrations may now accumulate to concentrations exceeding statutory levels.

During the same time, new knowledge has been accumulated for a number of indoor air quality (IAQ) parameters, e.g., carbon dioxide,¹ volatile organic compounds (VOC),² particulate matter³ and reactions between ozone and VOCs.⁴ This has, and will even more in the future, increase the restrictions on tolerable indoor air quality, not only in submarines, but in indoor environments in general.

A submarine is not only used during times of war, but also as a workplace and home during crew training. Thus, people working in submarines have, contradictory to people employed in most indoor environments, little changes in indoor air quality between working and leisure hours while in service.

Occupational exposure limits (OELs) are stated for 8 h continuous exposure, 40 h per week, during a period of 45 years. A typical submariner can be exposed to his or her working environment during a much longer continuous period, but for a shorter total time. The typical submariner can be assumed to be in on-board service 60–70 days per year during a period of 15 to 30 years. This implies that legislative levels are not applicable to the closed submarine environment. It is therefore important to establish OELs valid for the special conditions encountered in a submarine. This issue has been discussed before, e.g. by Raffaelli,⁵ and for the related space industry, OELs are already stated in USA⁶ and Europe,⁷ A study at the Swedish National Institute for Working Life (NIWL) have proposed guidelines for setting appropriate OELs for a number of compounds more or less likely to be present in submarine atmosphere.⁸ The data collected in our study can help determining today’s state-of-the-art IAQ in submarines and start to provide knowledge needed to ascertain appropriate OELs in the future.

Current knowledge of air quality in AIP submarines is rather sparse. This is especially true for extended times of submersion, i.e. longer than one week. The only published measurements so far have been performed by Swedish Defence Research Agency (FOI) during sea acceptance tests with the Gotland class submarines.⁹ The aim of the present study is to collect data for a number of air contaminants during eight days of continuous submersion.

Methods

The measurements were performed at winter conditions on-board a Gotland Class submarine during normal operation. The Gotland Class submarines are equipped with an air renewal system for the most obvious air contaminants. Carbon dioxide is removed with a soda lime based absorption system. Carbon dioxide and soda lime reacts as air is drawn through the soda lime bed. When the soda lime is saturated with carbon dioxide, it is replaced. The boats are also equipped with a catalyst in which hydrogen and carbon monoxide are oxidized. Furthermore, the boats are equipped with particle and activated carbon filters. The boats can operate with diesel engines in surface and snorting conditions and with Stirling engines and batteries in submerged condition. In this paper, no difference is made between Stirling and battery operation. The instruments were placed on-board at given positions and then supervised by a dedicated crew-member. The crew-member was trained on the functionality of the instruments and was given written instructions as well as a special logbook for each instrument.

The following compounds and pollutants were monitored within the study: carbon dioxide (CO₂), oxygen (O₂), ozone (O₃), hydrogen (H₂), nitrogen dioxide (NO₂), VOC, formaldehyde, airborne particulate matter (PM) and microbiological contaminants (represented by microbial markers: 3-hydroxy fatty acids, muramic acid and ergosterol). In addition, the physical properties of the air, i.e., temperature, total pressure (p_tot), and relative humidity, were monitored. After collecting the data, effort was made to correlate the air quality parameters, mostly for the particle measurements, to different sources and activities on-board.

During the trials, measurements were performed during 28 h air dependent operation, i.e., diesel operation, and 233 h air independent operation, i.e., battery and Stirling engine operation. The longest continuous operation in submerged condition during this time was 185 h, i.e. 7 days and 17 hours. To indicate were the different measurement equipment were located, the room configuration of the Gotland class submarines is shown in Fig. 1. The operational profile of the submarine and the sampling intervals for the measurement equipment during the campaign are shown in Fig. 2. Note that the boat was in harbour through days 3 to 7. The measurements were performed with both online instruments with high time-resolution (real-time measurements) and off-line integrated methods.

Fig. 1 Cross-sectional view of a Gotland class submarine.

Fig. 2 Operational profile of the submarine and the sampling intervals for the measurement equipment during the 15 day long measurement campaign. During days 3 to 7 the boat was in harbour. The lower value indicates Stirling/battery operation and the higher value diesel operation.

Real-time measurements

The real-time instruments used in the measurement campaign and their location on-board are presented in this section (Table 1). The real-time instruments were used to measure carbon dioxide, oxygen, ozone, hydrogen and VOC concentrations, airborne particulate matter, atmospheric pressure, temperature and relative humidity.

Table 1 Overview of real-time instruments used in the trials

Instrument	Parameter	Location	Logging interval	Range
a Toluene equivalents. dae = aerodynamic diameter.
AQM	O2	CR/SS	1 min^−1	0.01–100 kPa
	CO2	CR/SS	1 min^−1	0.01–5 kPa
	H2	CR/SS	1 min^−1	0.01–3%
	TVOC	SS	1 min^−1	0–200 ppm^a
	p tot	CR/SS	1 min^−1	0–140 kPa
	RH	CR/SS	1 min^−1	0–100%
	Temp	CR/SS	1 min^−1	0–100 °C
Analox	O2	EC/SS	2 h^−1	0–200 kPa
Hyperbaric	CO2	EC/SS	2 h^−1	0.01–10 kPa
Monitor	P atm	EC/SS	2 h^−1	1–10 bar
	Temp	EC/SS	2 h^−1	0–40 °C
Ship’s system	CO2	AMS	0.5 h^−1	0–5%
Ozone monitor	O3	CR	12 h^−1	0.0015–100 ppm
Optical particle counter dae 0.3–10 μm	Particle size distribution and number concentration	CR	1 min^−1	<350 particles/cm^3
Photometer dp < 10 μm	Particle mass concentration (PM10)	CR	30 h^−1	0.001–100 mg m^−3

---

Air quality monitor (AQM). The AQMs are autonomous air monitoring devices assembled by FOI.¹⁰ Two similar AQMs were used in the trials. The monitors were designed for air profile measurements in confined spaces. The monitors consist of a data-logger (AAC-2, INTAB Interface-Teknik AB, Sweden) and commercially available sensors for oxygen, carbon dioxide, hydrogen, total volatile organic compounds (TVOC) (only one of the AQMs), pressure, temperature and relative humidity. The TVOC sensor is sensitive to hydrocarbons with up to ten carbon molecules and ionisation energy below 10.6 eV, e.g., acetone, benzene, heptane, hexane, toluene and xylene. The sensor gives the TVOC concentration in toluene equivalents. The sensors were calibrated before and after the trials. Furthermore, the gas sensors were exposed to known gas concentrations at three times during the trials. During the trials, the AQM without TVOC sensor was located in the control room and the AQM with TVOC sensor was placed in the Stirling section.

Analox hyperbaric monitor. Analox SubMk II P (Analox Sensor Technology Ltd, Stokesley, Great Britain) is a portable analysis instrument designed for use in pressurised environments, e.g., a disabled submarine or a pressure chamber. Besides oxygen and carbon dioxide partial pressures, the instrument measures total pressure and temperature. The instrument was factory calibrated before and after the measurement campaign. The instrument was in use from the seventh day of the trials and placed in the electronic central. At 14:00 on the eleventh day, it was moved to the lower part of the Stirling section.

Ship’s internal system. The boat is equipped with an internal air monitoring system for particular gases (Dräger AG, Lübeck, Germany). The boat is equipped with sensors for oxygen, carbon dioxide, hydrogen and VOC. Only the carbon dioxide (Dräger Polytron) concentration measured with the ship’s system will be presented in this study. The calibration status of the ship’s system is not known. The carbon dioxide concentration sample was taken from the auxiliary machinery space.

Ozone monitor. An UV absorption based instrument was used for ozone monitoring (Ozone Monitor Model 202, 2B Technologies Inc., Golden, Colorado, USA). The instrument monitors ozone by alternatively measuring the absorption of light at 254 nm from an air sample drawn directly to the sensor, and air first drawn through an ozone scrubber (2BTech 2001). The ozone concentration is obtained by comparing the two measured absorptions. The detection range of the instrument is 0.0015 to 100 ppm. The instrument was calibrated in November 2002.

During the trials, the instrument was placed beside the steering panel on the port side of the control room. The sampling inlet was located 20 cm above the floor plate. The ozone concentration was analysed continuously and time weighted averages were stored in the instrument’s internal memory.

Optical particle counter (OPC). An optical particle counter, (Climet CI-500, Climet Instruments Inc., Redlands, California, USA) was used to measure the particle size distribution on-board. The OPC divides the number of particles in six particle sizes, 0.3–0.5, 0.5–1, 1–5, 5–10, 10–25 and >25 μm, based on the intensity of the scattered light for individual particles. In this work, only particles below 10 μm are considered. The instrument was calibrated regarding size and concentration measurements prior to the measurement campaign. Monodisperse di-ethyl-hexyl sebacate aerosol (a common liquid calibration aerosol) was used. The detection efficiency from comparisons with a calibrated condensation particle counter was 30% for particles with an aerodynamic diameter of 0.3 μm and 90% for 0.5 μm, respectively. The upper concentration limit is 350 particles per cm³, above this level recorded particle concentrations are underestimated due to coincidence in the sensor.

During the trials, the instrument was placed beside the steering panel on the port side of the control room, next to the ozone monitor. The sampling inlet was located 30 cm above the floor plate. The sampling flow-rate was checked on a daily basis. The instrument measured the particle concentrations continuously and time weighted averages were stored on a laptop computer, using a purpose developed Labview^TM-program (National Instruments Corporation, Austin, Texas, USA).

Photometer. Besides the optical particle counter, a light scattering laser photometer was used for particulate matter (PM) measurements (DustTrak 8520, TSI inc, Minnesota, USA). The mass concentration of PM is estimated from the total light-scattering intensity of particles present in the sensor. The DustTrak was equipped with an inlet impactor with an aerodynamic cut-off diameter of 10 μm for assessment of PM₁₀. Before the trials, the instrument had been factory calibrated with “ISO 12103-1 A1 Arizona Test Dust”.

The instrument was placed beside the steering panel, on the port side of the control room, next to the ozone monitor and optical particle counter. The sampling inlet was located 40 cm above the floor plate.

Off-line measurements

The methods used in the measurement campaign and their location on-board are presented in the section below. VOC sampling was performed with active sampling using adsorption onto Tenax tubes and charcoal tubes, and passive sampling using Tenax adsorption tubes and personal dosimeters. To estimate the total amount of volatile organic compounds (TVOC) all components were given as Toluene equivalents. Note that TVOC is an unspecified value used to describe the quality of indoor air. In each sample the different compounds must be considered. Furthermore, formaldehyde and nitrogen dioxide were measured with diffusive methods. Microbiological contamination was determined from dust and water samples taken before and after the campaign.

Stationary active sampling of VOC. Both Tenax TA (Perkin Elmer) and charcoal adsorbent tubes (Anasorb CSC Lot 2000, SKC Anasorb CSC Coconut Charcoal 226-01) were used for VOC sampling. The tubes were connected to pumps, which were calibrated before and after the campaign. Typically, 3 to 4 l was drawn through each tube for both the Tenax and charcoal tubes during a period of one hour. During the trials, all actively sampled VOC measurements were taken from the control room.

After the trials, the Tenax tubes were analyzed using gas chromatography and mass spectrometry (GC-MS) (GC Model 8500, ATD-400, Perkin Elmer Ltd, MA, USA, Ion Trap 800, Thermo Electron Corporation, Woburn, MA, USA) by Chemik Lab AB in Norrtälje, Sweden. The charcoal tubes were analyzed at the School of Engineering, Kristianstad University, Kristianstad, Sweden. The samples were desorbed using carbon disulfide and analysed using GC-MS (trace GC, Quadrupole MS type SSQ700, Thermo Electron Corporation, Woburn, MA, USA).

Personal passive sampling of VOC. Personal exposure to VOCs was evaluated using passive sampling with personal dosimeters (SKC 575–002 with 500 mg Anasorb 747, SKC Inc., PA, USA) and Tenax TA diffusion tubes. The dosimeters and tubes were attached to the persons left breast-pocket and adsorbed VOCs by diffusion.

In the trials, six dosimeters were used. The sampling times of the dosimeters are given in Fig. 2. The dosimeters were analysed with GC-MS in the same way and by the same laboratory as described above for the pumped charcoal tubes.

In the trials, six tubes were used. The sampling times of the dosimeters are viewed in Fig. 2. The Tenax TA diffusion tubes were analysed with GC-MS in the same way as for the actively sampled Tenax tubes.

Personal passive sampling of formaldehyde. Personal dosimeters (UME^X 100, Scantec, Gothenburg, Sweden) were used for formaldehyde sampling. Formaldehyde in the air diffuses to a filter impregnated with 2,4-dinitrofenylhydrazin and retrospectively, the reaction product was desorbed and analysed using liquid chromatography and UV detection by the Occupational and Environmental Medicine, Sahlgrenska University Hospital, Gothenburg, Sweden.

In the trials, twelve dosimeters were used. Eight of the dosimeters were used in submerged condition. The eight dosimeters were exposed four and four at two separate times. These dosimeters were exposed to the submarine atmosphere for approximately three days. Four of the dosimeters were worn by crew-members in the forward section and the other four were worn by crew-members in the aft section. One of the dosimeters was lost during the trials. The final four dosimeters were exposed during snorting operation for approximately nine hours. Two of the dosimeters were worn by crew-members in the forward and two by crew-members in the aft section.

Personal passive sampling of nitrogen dioxide. Nitrogen dioxide sampling was performed using passive diffusive samplers (IVL, Gothenburg, Sweden). After the trials, the samplers were analysed by IVL using spectrophotometer and flow injection analysis.

Six samplers were used in the trials. Four of the samplers were used in submerged condition. The four samplers were exposed two and two at two separate times. These samplers were exposed to the submarine atmosphere for approximately three days. Two of the samplers were worn by crew-members in the control room and the other two were worn by crew-members in the electronic central. The final two samplers were exposed during snorting operation for approximately nine hours. One of these samplers was placed in the control room section and the other in the electronic central.

Microbiological measurement. Samples for microbiological analysis were taken as dust from the filter cloth in the central air conditioning system, and as water from the bilge in the auxiliary machinery space, in the forward part of the Stirling section and in the machinery room. The samples were analysed using GC-MS to determine and quantify biomarker molecules.¹¹

Microorganisms contain unique compounds, which can be used as chemical markers of larger, bioactive structures. Endotoxins (LPS) are major constituents of the outer membrane of gram-negative bacteria. A backbone of Lipid A, the toxic component of the LPS molecule, carries in general 4 moles of unique 3-hydroxy fatty acids of carbon-chain lengths 10 to 16. Muramic acid is a unique marker of peptidoglycan, which is mostly found in gram-positive bacteria and in small amounts in gram-negative bacteria. Ergosterol is a common fungal membrane lipid that is widely used as a marker of fungal biomass.

The method requires a dust sample of 1–5 mg or a freeze-dried water sample of approximately 1 ml. Briefly, the samples were hydrolyzed to release markers, purified using silica gel columns, derivatised, and analyzed in electron impact mode using an ion-trap GC-MS (Saturn 2000, Varian, Palo Alto, CA, USA) equipped with a fused-silica capillary column (Chrompack, Middelburg, The Netherlands).

Results

Pressure

The total pressure (results from pressure measurements AQM) varied between 92 and 106 kPa and highly depended on the operational profile and the activities on-board (Fig. S1, ESI†). In diesel mode, the pressure decreases since the diesel engines draw air from the outside of the boat. The average pressure was approximately 4 kPa lower in diesel operation compared with AIP operation. In AIP operation, when the boat is sealed, the pressure increased due to a leakage in the high pressure air system. Soda lime exchange is performed using pressurized air. Due to this, an increase in pressure is observed in conjunction with the soda lime exchanges. To reduce the pressure to normal levels, the high-pressure compressor is run either simultaneously to the soda lime exchange or shortly after each exchange. The variation in pressure increase between the exchanges is due to differences in time between the exchange and start of the compressor.

Temperature and relative humidity

The temperature in the control room varied between 17 and 24 °C (Fig. S1, ESI). In the Stirling section, the temperature varied between 13 and 22 °C. The reason for the difference is that most personnel and electronic equipment are located in the control room. The liquid oxygen tanks, located in the Stirling section also contribute to the lower temperature observed there.

In the control room, the relative humidity varied between 38 and 71%. In the Stirling section, the relative humidity varied between 42 and 70%.

Oxygen

The oxygen concentration varied between 18.5 and 21% (Fig. S2, ESI†). The instrument measured partial pressure, which has been recalculated to volume fractions (%). The concentration in the control room and Stirling section were similar in level and profile. During AIP operation, consumed oxygen was replenished from the LOX tanks. As the oxygen supply was turned on, the concentration remained at a constant level. During days eight and nine, a decrease in oxygen concentration is observed. The reason for this is that the oxygen supply was not started until day nine at 18.15. Before this time, oxygen was only consumed, not added to the atmosphere.

Carbon dioxide

Results from measurement of the carbon dioxide concentration (AQM) in the control room are shown in Fig. 3. During the long AIP mode period, i.e., days eight through fifteen, the mean carbon dioxide concentration in the control room and Stirling section was 0.48 and 0.56%, respectively. In the control room, the carbon dioxide concentration varied between 0.00 and 0.70% and in the Stirling section it varied between 0.10 and 0.65%.

Fig. 3 Carbon dioxide concentration in the control room and Stirling section (AQM) as a function of time and operational profile. Soda lime exchanges are marked with triangles. Carbon dioxide absorption is indicated with the thick grey line.

The current Swedish 8 h OEL for carbon dioxide is 0.5%.¹² This is the same concentration as the design carbon dioxide concentration for the carbon dioxide removal system on the Gotland Class submarines. The proposed maximum concentration of carbon dioxide according to SNIWL⁸ and SMAC_{30 day}⁶ is 0.5% and 0.7%, respectively.

In the figure, a sudden increase in carbon dioxide concentration is observed as break-through appears in the soda lime, i.e., as the soda lime is saturated with carbon dioxide. Similarly, a decrease in carbon dioxide concentration is observed after each soda lime exchange.

Fig. 4 shows the carbon dioxide concentration in the control room (AQM), Stirling section (AQM), lower part of the Stirling section (Analox Hyperbaric monitor) and the auxiliary machinery space (Ship’s system), during day eight of the measurement period. Absolute comparison of the values is not possible since the calibration of the ship’s system, i.e., measuring in the auxiliary machinery space, could not be controlled. The figure shows the decrease in carbon dioxide concentration as diesel operation is begun, i.e., as the boat is ventilated against the atmosphere. The concentration in the control room decreases rapidly in the beginning. After approximately 2 h, the concentration in the control room reaches ambient level. In the Stirling section, the concentration decreases at a more moderate rate. Ambient level is not reached even at the end of the diesel operation, i.e., after 4 h. This indicates that the time needed to fully ventilate the boat is considerable.

Fig. 4 Carbon dioxide concentration in the control room (AQM), Stirling section (AQM), lower part of the Stirling section (Analox Hyperbaric monitor) and in the auxiliary machinery space (ship’s system), during day eight of the measurement campaign. This day incorporated diesel as well as AIP operation. Soda lime exchange is indicated with a triangle and carbon dioxide absorption with the thick grey line.

Hydrogen

During the long AIP mode period, i.e., days eight through fifteen, the mean hydrogen concentration in the control room and Stirling section was 0.3 and 0.2%, respectively (Fig. S3, ESI†). The increase in hydrogen concentration between days eight to eleven is due to faster cruising speed of the boat, resulting in an increase in battery output and thus also in hydrogen production on-board. The peak at the end of day sixteen is due to the same cause. The lower explosion level (LEL) of hydrogen is 4.1%, i.e., the concentrations reached in the boat are well below this value. The proposed maximum concentration of hydrogen according to SNIWL⁸ and SMAC_{30 day}⁶ is 0.4% and 0.41%, respectively.

Volatile organic compounds

The measured and calculated TVOC concentrations are all presented as toluene equivalents. In some cases, specific compounds within the samples have been identified and quantified.

The TVOC concentrations calculated from analysis of actively sampled Tenax TA and charcoal tubes are given in Table 2. Before the first samples were taken, the diesel engines were running for 3 hours and 45 minutes. The engines then continued running during the sampling time. The other samples were taken when the boat was operated in continuous AIP mode. In accordance with the results obtained with the AQM, no increase in TVOC can be observed with increased time of submersion from neither Tenax TA nor charcoal tube analysis.

Table 2 TVOC results from actively sampled Tenax TA and charcoal tubes expressed as mg m⁻³ Toluene equivalents

Start time	End time	Mode	Tenax	Charcoal
Day 8 18:57	Day 8 19:58	Diesel	0.6	1.6
Day 9 19:23	Day 9 20:22	AIP	0.7	1.1
Day 10 18:50	Day 10 20:01	AIP	0.7	0.7
Day 11 19:43	Day 11 20:51	AIP	0.5	0.8
Day 12 18:48	Day 12 19:50	AIP	0.8	0.8
Day 13 18:40	Day 13 19:46	AIP	0.6	0.8
Day 14 18:34	Day 14 19:35	AIP	0.7	0.9
Day 15 19:12	Day 15 20:13	AIP	0.9	0.8

---

The TVOC concentrations calculated from the analysis of the dosimeters are shown in Table 3. The first dosimeter in the table was, due to unknown reason, not exposed during the trials. No significant differences can be observed between the different samples.

Table 3 TVOC results from personal dosimeter analysis expressed as mg m⁻³ toluene equivalents

Start time	End time	Mode	Section	TVOC conc/mg m^−3 toluene equiv.
Day 13 23:00	Day 14 05:02	AIP	Aft	Not exposed
Day 13 23:00	Day 14 05:02	AIP	Aft	0.2
Day 13 23:00	Day 14 05:02	AIP	Forward	0.3
Day 13 23:00	Day 14 05:02	AIP	Forward	0.1
Day 16 00:00	Day 16 09:00	Diesel	Aft	0.2
Day 16 00:00	Day 16 09:00	Diesel	Forward	0.1

---

The TVOC concentrations calculated from the Tenax TA diffusion tubes are shown in Table 4. The third diffusion tube in the table could not be analysed properly due to malfunction of the GC-MS. No significant differences can be observed between the forward and aft sections. Similarly, no significant difference is observed between AIP and diesel operation.

Table 4 TVOC results from Tenax diffusion tube analysis expressed as mg m⁻³ toluene equivalents

Start time	End time	Mode	Section	TVOC conc/mg m^−3 toluene equiv.
Day 8 12:30	Day 11 12:30	AIP	Aft	1.5
Day 11 12:30	Day 14 12:30	AIP	Aft	1.7
Day 8 12:30	Day 11 12:30	AIP	Forward	—
Day 11 12:30	Day 14 12:30	AIP	Forward	0.9
Day 16 00:00	Day 16 09:10	Diesel	Aft	1.2
Day 16 00:00	Day 16 09:10	Diesel	Forward	2.4

---

Fig. 5 shows the ten VOCs that were identified in highest concentrations in the charcoal tube analysis. These substances were identified with accuracy above 70%. In total, 49 compounds could be identified with accuracy above 70%. The identified compounds contribute to approximately 10% of the TVOC concentration shown in Table 2. The results show no significant changes in concentration of the identified compounds with increased time of submersion. Today, no Swedish OEL is stated for TVOC concentration.

Fig. 5 The ten compounds, present in highest concentration, that could be identified, with an accuracy above 70% with retrospective charcoal tube analysis, as a function of time of submersion. The concentrations for all compounds are shown in toluene equivalents. In the figure, the compounds are sorted after the maximum concentrations found in the tube taken day 8. In total, 49 compounds were identified with an accuracy above 70%. The identified compounds contribute to approximately 10% of the TVOC concentration.

Formaldehyde

The formaldehyde measurements showed concentrations below 0.01 mg m⁻³. The current Swedish 8 h OEL for formaldehyde is 0.6 mg m⁻³.¹² The “Generation target level” specified by the Swedish Environmental Protection Agency (SEPA) is 0.01 mg m⁻³.¹³ The proposed maximum concentrations of formaldehyde according to SNIWL,⁸ SMAC_{30 day}⁶ and ESA_{10 day}⁷ are 0.125, 0.05 and 0.36 mg m⁻³, respectively.

Nitrogen dioxide

The nitrogen dioxide measurements showed concentrations below 0.02 mg m⁻³. The current Swedish 8 h OEL for nitrogen dioxide from engines is 2 mg m⁻³.¹² A better comparison is the new 98 percentile target value for outdoor nitrogen dioxide suggested by the Swedish EPA. The suggested target value, based on one hour averages, is 0.06 mg m⁻³.¹³

Ozone

The only conclusion that can be drawn from the ozone measurement is thus that the background level is low (<5 ppb). The current 8 h OEL in Sweden is 100 ppb.¹² The “Generation target level” of ozone during summer, specified by the Swedish EPA, is 25 ppb. The proposed maximum concentration of ozone according to both SNIWL⁸ and ESA_{10 day}⁷ is 20 ppb. The peaks in the measured data correlate strongly to the measured particulate matter peaks (Fig. S4 and S5, ESI†). Recent findings⁴ have indicated a connection between ozone and particle concentration through gas to particle conversion from chemical reactions of ozone and VOCs (e.g. terpenes). However, due to the measurement principle of the ozone instrument, it is likely that the instrument is cross-sensitive to particles. This means that the measured ozone peaks may not be due to ozone but rather interference from particulate matter. Particles may be preferentially lost in the ozone scrubber and thus contribute to the differential light absorption signal. The manufacturer of the ozone monitor supports this hypothesis.¹⁴

Particulate matter

Measured particle number concentrations (OPC) and assessed PM₁₀ mass (photometer) concentrations in the control room are shown in Fig. 6–8 (Fig. S5, ESI†). As can be seen from Fig. 7, particles between 0.1 and 1 μm have, as can be expected, the strongest contribution to the total number concentration. The results from the different particle size intervals coincide for some of the peaks, but not for all. This implies that there are a number of sources on-board generating high particle concentrations. In some peaks, the number of particles detected exceeds the instrument’s upper concentration limit, i.e., the actual particle concentration is even higher than detected. Due to the high air exchange rate in the control room, the generated particles are quickly removed as the particular source is disabled. The time for the concentration to resume background level is approximately 30 minutes. Fig. 7 shows results for particles between 0.3 and 1 μm during part of days seven and eight. Sources that have been identified using the ship’s logbook are indicated with arrows. The mast operation peak can be seen as typical as it was of similar appearance each time a mast was operated. The peak correlating with soda lime exchange is not to be seen as typical. The peak height varied for different soda lime exchanges. The variation in peak height might be caused by the operational behaviour of the crew-member performing the exchange. Similar to the soda lime exchange peaks, the diesel peak heights varied between diesel operations.

Fig. 6 Particle number concentration (optical particle counter) in the control room for particles between 1 and 10 μm as a function of time and operational profile. Soda lime exchanges are marked with triangles.

Fig. 7 Particle number concentration (optical particle counter) in the control room for particles between 0.3 and 1 μm during days seven and eight. The upper concentration limit of the optical particle counter is indicated by the bold black line, therefore, the data above 350 particles per cm³ is only a lower limit for the true concentration. Soda lime exchanges are marked with triangles. Probable sources to increased concentration are indicated with arrows.

Fig. 8 PM₁₀ concentration (photometer) in the control room as a function of time and operational profile. Soda lime exchanges are marked with triangles.

Data obtained with the photometer, see Fig. 8, correlates well with results from the optical particle counter, see Fig. 6–8 (Fig. S5, ESI†), i.e., with certain activities on-board, strong short-lived particle peaks can be observed. The current 8 h OELs in Sweden are for respirable dust 5 mg m⁻³ and for textile dust 1 mg m⁻³. The mean PM₁₀ value obtained during the whole period of these trials, i.e., from day one to sixteen is 0.03 mg m⁻³. As comparisons, concentrations of particles expressed as PM₁₀ are 0.008–0.017 mg m⁻³ in rural areas in Sweden, 0.014–0.023 mg m⁻³ in urban background air, and 0.035–0.044 mg m⁻³ in the street.¹³

Microbiological contamination

The results from the assessment of microbial flora are shown in Table 5. The concentrations of chemical markers representing gram-positive bacteria, gram-negative bacteria and fungi in the dust sample resemble concentrations found previously in home environments^11,15 and work environments,^16,17 using the same technique.

Table 5 Amounts of lipopolysaccharide (LPS), muramic acid and ergosterol in airborne dust samples and three different locations in bilge before and after the mission

Sample	Lipopolysacharide/nmol mg^−1 dust or ml water	Muramic acid/ng mg^−1 dust or ml water	Ergosterol/ng mg^−1 dust or ml water
Airborne dust after the mission/mg	0.12	5.4	0.2
Bilge HMAR before the mission/ml	0.48	71.1	35.7
Bilge HMAR after the mission/ml	0.66	85.1	11.1
Bilge SR before the mission/ml	0.20	11.9	2.7
Bilge SR after the mission/ml	0.76	52.9	1.9
Bilage MAR before the mission/ml	0.05	9.4	4.8
Bilage MAR after the mission/ml	0.34	65.8	13.2

---

The marker levels in water samples show tendency to increase during the campaign, with the exception of ergosterol. The reason why ergosterol does not follow the pattern cannot be readily determined. We also found quite high relative amounts of 3-hydroxy fatty acids of 10 and 12 carbon chain length (data not shown), which may suggest greater presence of gram-negative bacteria. Usual levels of these fatty acids in the air imply that microorganisms are not easily airborne in these specific conditions.

Discussion

The data obtained in this work imply that no accumulation of the monitored air contaminants occur during eight day long submerged missions as long as the air renewal systems work properly.

The pressure within the submarine varies around normal atmospheric pressure with peaks up to 105 kPa. By using the on-board compressor, pressure variations due to consumption from the pneumatic system can be equalized.

Oxygen is maintained at acceptable level by the crew during the entire mission. Due to the AIP system and the large amounts of oxygen stored for energy conversion in submerged condition, oxygen is available in amounts far larger than required for the crew metabolism.

During the extended AIP period, the temperature in the control room levelled off at approximately 24 °C at a position approximately 1.8 m above the platform. Relative humidity is kept at approximately 55%. Due to the nature of the submarine, i.e., high heat loads within a crowded confined volume, thermal conditions are hard to optimise. However, by implementing a better thermal climate in the occupied spaces, it is possible that a better perceived air quality could be obtained.

At the end of the long AIP period, the 8 h average of the carbon dioxide concentration exceeds 0.5%. This rise could probably have been avoided if the soda lime exchanges had been performed slightly earlier than they actually were. As the soda lime is saturated with carbon dioxide, a sudden increase in carbon dioxide concentration is observed. The current soda lime exchange criterion is based on the time since the last exchange and progress of the carbon dioxide concentration. By monitoring the outlet concentration from the soda lime, the time of break-through may be seen as soon as it appears and thus a better change criterion for the soda lime can be applied. It should be noted that an earlier change of soda lime is likely to result in decreased absorption capacity, resulting in decreased submerged operational radius. An association between increased carbon dioxide concentration and sick building syndrome symptoms has been observed in the literature. However, carbon dioxide has been found to be approximately correlated with other indoor pollutants more likely to cause symptoms.¹ Increased carbon dioxide concentrations may therefore be accepted in a submarine without endangering the health of the crew.

The hydrogen concentration in the occupied compartments is kept below one tenth of the lower explosion limit (LEL), i.e., well below any explosion risk. It should be noted that within the battery compartments, local concentrations above LEL are unavoidable due to the construction of the battery cells.

Particulate matter concentrations are well below any 8 h OELs and is within the same range as in outdoor urban air. The PM₁₀ levels are comparable to levels given by the Swedish EPA.¹³ Thus, it can be concluded that the particle filter units in the ventilation system on-board seem to work satisfactorily. A photometer could probably be used to detect sources of particle contamination. By identifying sources, particle formation may perhaps be prevented in new constructions or refits, resulting in removal of the strong short-lived peaks detected in this study. The sources giving rise to particulate matter are likely to generate other air contaminants, e.g., VOCs and SVOCs as well, i.e., the detection, and later removal, of contaminant sources may contribute to more than only decreased particulate matter concentration. The detected pattern resembles findings obtained using the same type of photometer on-board diesel-electric submarines by Mazurek and Gan¹⁸ They pointed out textile fibres, cooking and diesel exhausts as the main contributing sources of airborne particles on-board. Morawska et al.¹⁹ have measured the mass concentrations in residential houses with the same type of photometer, but with a 2.5 μm impactor mounted on the instrument, i.e., the instrument measured PM_2.5. The average mass concentration found in their study was between 0.01 and 0.015 mg m⁻³. Even though the results are not directly comparable, since the instruments have not measured the same size intervals, it can be concluded that both measurements are within the same range.

The VOC measurements show a low total concentration. An increase in VOC concentrations with increased diving time cannot be seen during the eight-day period. This indicates that the charcoal filter installation can adsorb the substances emitted from various sources. By combining the results from the charcoal tube analysis with results from GC-MS analysis of the hydraulic and diesel oils used on-board, it seems likely that these systems, together with lubricating oils, contribute largely to the overall VOC load. To determine the actual conditions, more detailed analysis of the chromatograms is required than performed so far. Seifert²⁰ presents guidelines for TVOC based on a certain TVOC mixture, but the results are not directly comparable to the results from this study due to differences in measurement technique and analyzed substances. In his study, Seifert states that a TVOC concentration below 0.2 mg m⁻³ does not give any noticeable effect on people and that a concentration in the range 0.2 to 3 mg m⁻³ possibly may give rise to irritation or decreased well-being in combination with other exposure parameters. The concentrations detected within this study are within the 0.2–3 mg m⁻³ range, and thus fall within Seifert’s second range.

Both the nitrogen dioxide and the formaldehyde measurements showed low concentrations. Nitrogen dioxide was found at 0.02 mg m⁻³, which can be compared to the current eight hour OEL of 2 mg m⁻³. Formaldehyde was found in low concentration on-board, i.e., 0.01 mg m⁻³, which can be compared to the eight hour OEL presently stated for formaldehyde in Sweden, i.e., 0.6 mg m⁻³.

Apart from carbon dioxide, all measured concentrations are well below current 8 h OELs¹² were such values are available, and at or below values specified by the Swedish EPA¹³ for outdoor air, SNIWL,⁸ NASA⁶ and ESA.⁷ The determination of submarine specific OELs is a hard task. Even within “normal” IAQ research, for example in residential buildings, with comparably large research funding such levels are hard to determine. As an example, Moelhave²¹ concludes that further research is required before an OEL can be set for TVOC concentration. However, it should be kept in mind that only a few of the contaminants present in submarine environment were actually investigated. It can be assumed that a substantial build-up of the investigated contaminants is not likely if the submersion period is multiplied, which is the case for modern submarines equipped with air independent propulsion.

Acknowledgements

Kockums A. B. and the Swedish Knowledge Foundation are gratefully acknowledged for their financial support of this project. Thanks also to Rickard Pellny and the rest of the crew on HSwMS Halland for taking excellent care of the equipment during the measurement campaigns. Mårten Spanne, Eddie Bergsten, Diauddin Nammari and Tina Hjellström are gratefully acknowledged for their efforts and interest in the project.

References

1. M. Apte, W. Fisk and J. M. Daisey, Associations between indoor CO₂ concentrations and sick building syndrome symptoms in US office buildings: an analysis of the 1994–1996 BASE study data, Indoor Air, 2000, 10, 246–257.
2. K. Andersson, J. V. Bakke, O. Bjørseth, C.-G. Bornehag, G. Clausen, J. K. Hongslo, M. kjellman, S. Kjærgaard, F. Levy, L. Mølhave, S. Skerfving and J. Sundell, TVOC and health in non-industrial indoor environments, Indoor Air, 1997, 7, 78–91.
3. T. Schneider, J. Sundell, W. Bischof, M. Bohgard, J. W. Cherrie, P. A. Clausen, S. Dreborg, J. Kildesø, S. K. Kjærgaard, M. Løvik, P. Pasanen and K. Skyberg, EUROPART. Airborne particles in the indoor environment. A European interdisciplinary review of scientific evidence on associations between exposure to particles in buildings and health effects, Indoor Air, 2003, 13, 38–48.
4. C. Weschler, Ozone in indoor environments: concentrations and chemistry, Indoor Air, 2000, 10, 269–288.
5. P. Raffaelli, Occupational hygiene problems in submarines-OELs for continuous exposure for prolonged periods of up to 3 months, Ann. Occup. Hyg., 1989, 33(1), 117–121.
6. NASA, Spacecraft Maximum Allowable Concentrations for Selected Airborne Contaminants, PB 95-106068, NASA, Washington, DC, USA, 1994.
7. ESA, Atmosphere Quality Standards in Manned Space Vehicles, ESA PSS-03-401, European Space Agency, Noordwijk, The Netherlands, 1992.
8. A. Filipsson and G. Johanson, Continuous exposure to air contaminants during long periods, in Swedish Kontinuerlig exponering för luftföroreningar under lång tid, Arbetslivsinstitutet Enheten för Toxikologi och Kemi, 1997.
9. M. Loncar, Air samples onboard submarine Gotland, Luftprover ombord ubåten Gotland, FOA–R–97-00403-720–SE, FOA, Hårsfjärden, 1997, ISSN 1104-9154 (in Swedish).
10. J. Risberg, C. Ostberg, T. Svensson, W. Norfleet, H. Ornhagen, O. Mjaavatten and T. Juvik, Atmospheric changes and physiological responses during a 6-day “disabled submarine” exercise, Aviat. Space Environ. Med., 2004, 75, 138–149.
11. A. Sebastian and L. Larsson, Characterization of the microbial community in indoor environments: a chemical-analytical approach, Appl. Environ. Microbiol., 2003, 69, 3103–3109.
12. Swedish Work Environment Authority, AFS 2000:3, SWEA, 2000.
13. Swedish Environmental Protection Agency, Frisk Luft (Clean Air), Naturvårdsverket, 2003, p. 5318 (in Swedish).
14. J. Birks, Lund Institute of Technology, Lund, 2004, E-mail: ola.persson@vok.lth.se.
15. A. Saraf, L. Larsson, H. Burge and D. Milton, Quantification of ergosterol and 3-hydroxy fatty acids in settled house dust by gas chromatography-mass spectrometry: comparison with fungal culture and determination of endotoxin by a Limulus amebocyte lysate assay, Appl. Environ. Microbiol., 1997, 63(7), 2554–2559.
16. A. Fox, W. Harley, C. Feigley, D. Salzberg, A. Sebastian and L. Larsson, Increased levels of bacterial markers and CO₂ in occupied school rooms, J. Environ. Monit., 2003, 5(2), 246–252.
17. C. J. Hines, M. A. Waters, L. Larsson, M. R. Petersen, A. Saraf and D. K. Milton, Characterization of endotoxin and 3-hydroxy fatty acid levels in air and settled dust from commercial aircraft cabins, Indoor Air, 2003, 13(2), 166–173.
18. W. Mazurek and T. Gan, Airborne Particulates in Diesel-Electric Submarines, SAMAP, Emden, Germany, 2003.
19. L. Morawska, C. He, J. Hitchins, K. Mengersen and D. Gilbert, Characteristics of particle number and mass concentrations in residential houses in Brisbane, Australia, Atmos. Environ., 2003, 37, 4195–4203.
20. B. Seifert, Richtwerte für die Innenraumluft. Bundesgesundheitsbl-Gesundheitsforsch-Gesundheitsschutz, Springer-Verlag, 1999, pp. 270–278.
21. L. Moelhave, Organic compounds as indicators of air pollution, Indoor Air, 2003, 13, 12–19.

---

Footnote

† Electronic supplementary information (ESI) available: Fig. S1–S5, air quality monitoring. See DOI: 10.1039/b605331a

---