Oil-leakages on the 0vre Romerike aquifer, Southern Norway

GAUTE STORR0 & DAVID BANKS

Storre, G. & Banks, D. 1992 : Oil-leakages on the 0vre Romerike aquifer , Southern Norway. Nor. geol. unders. Bull 422, 67-81.

Two leakages of, respective ly 20 m' and < 10 m' of fuel oil have been investigated at Trandum and Sessvollmoen , on Norway 's 0vre Romerike aquifer. The hydrocarbon content of the unsatura­ ted and saturated sediments, and the hydrocarbon content of groundwater at both sites have been investigated. At Trandum , where the unsaturated zone consists largely of coarse sands, the oil appears to have migrated rapidly down towards the water table, and high dissolved hydrocar­ bon concentrations (0.6 - 1.6 mg/l) have been observed in groundwater from scavenger wells . At Sessvollmoen , the unsaturated zone consists of finer sediments , and the oil appears to have been retained at a relatively shallow depth , allowing only minor downwashing of soluble compo­ nents to the water table. The scavenged groundwater contains around 0.05 mg/l hydrocarbon.

The contrast between the two case studies illustrates the importance of the type of sediment in the unsaturated zone, and thus its retention capacity, when assessing the impact of oil spills on aquifers.

Gaute Storm & David Banks , Geological Survey of Norway , Postboks 3006 - Lade, N-7002 Trend­ tieim, Norway.

Introduction Areal use and potential pollution The 0vre Romer ike aquifer is Norway's lar­ sources cest discrete aquifer. It has a total area of The Romerike aquifer is not urbanised, but some 105 km' (Davidsen 1990) and is compo­ nevertheless is the host to several military sed of Quaternary deposits, largely of glacioflu­ bases , a military airstrip, a civil airport (Garder­ vial origin. At present, the aquifer is only light­ moen), and diverse military exercises. The ly exploited, but it has been the focus of area is also intensively worked for sand and much recent attention because of several pollu­ gravel. The Romerike deposit is believed to tion threats. Authorities have placed emphasis contain 150-200 million m' of good quality on protecting the quality of the groundwater sand and gravel (Wolden & Erichsen 1990). in the aquifer from «potentially polluting activi­ The aquifer contains Norway's largest discrete ties », as it represents a possible future water reserve of fresh groundwater, although at pre­ resource for municipalities in the Romerike sent it is only lightly exploited, supplying the area (0stlandskonsult et al. 1991). military bases and the local municipality's ne­ The area was the subject of one of Nor­ eds. Authorities have, however, emphasised way 's earliest attempts at environmental «Geo­ the need to preserve the quality of the aqui­ plan» mapping, where various user-interests, fer's groundwater for more extensive potential including groundwater interests, were conside­ future use in supplying the municipalities of red in the context of future areal planning the Romerike area . Wolden & Erichsen (1990) (Wolden & Erichsen 1990). have compiled «Geoplan» maps illustrating the It was in such a context that two oil leaka­ various sand and gravel, groundwater and ges at Trandum (20 m') and Sessvollmoen geological conservation interests which should « 1Orn') were discovered. These attracted be preserved in the area . much attention from the local and national There are at present several potential polluti­ media. The polluter (the Norwegian Defence on threats to the aquifer: Construction Service) was ordered by the Sta­ te Pollution Prevention Agency (SFT) to under­ (a) the existing civil & military airport at Garder­ take extensive investigations and to produce moen , and the proposed removal of Oslo 's plans for remedial action. It is the results of main airport thither (Davidsen 1990). Propylene these investigations that this paper addresses. glycol and urea are used as de-icing agents, 68 Gaute Storm & David Banks NGU - BULl. 422. 1992

Fig. 1. Map of Trancurn/Sessvotlrnoen area. showing hyd rogeolog ical features and variou s poss ible pollution threats. 0235-numbers refer to localities in Mor land et al. (19901.

and there is also the threat of major spillages marg inal delta sediments built up to the mari­ of oil or kerosene. ne limit. The upper part of the aquifer con­ (b) the numerous military bases & training sists of dom inantly glaciofluvial sand and gra­ grounds on the aquifer. At these large amounts vel depos its, with areas of aeolian sand and of oil and diesel are used for heating purpo­ glacio lacustrine sands and silts . These are ses and for refuelling vehicles. The possibil ity underlain by glaciomarine/mar ine silts and cannot be exclu ded that diesel has also been clays . The upper, coa rser part of the depo sit used for 'washing-down' of vehicles . exceeds 30-40 m thickness in some areas, (c) the recent national hazardous waste survey while the tota l depth to bedrock (including (Misund et al. 1991a.b) identified several poten­ marine silts and clays) may be as much as tially serious landfills and instances of contami­ 100 m (0 stmo 1976, Je rqense n & 0stmo 1990). nated ground (Morland et al. 1990). The main surface water dra inage of the (d) the discovery, in the latter part of '1990, aquifer consists of the northw ards-flowin g of two separate incidents involving the leakage River Risa and Hersjoen Lake . The River and of fuel oil from storage tanks , at Trandum and Lake are almost entirely groundwater-fed (J0r ­ Sessvollmoen military bases (Storre 1991, gensen & 0stmo 1990). Banks 1991). There has been intense hydrogeological and hydrochemical investigation of the aquifer due The locations of these pollution hazar ds are to its selection as a study area for the Interna­ shown on Fig.1. tional Hydrological Decade (Falkenmark 1972, Norwegian National Comm ittee for IHD 1973, 1975). This has resu lted in the publ ication of a hydrogeological map (0 stmo 1976), and Geology and hydrogeology descriptions of the hydrogeo logy (Jerqensen The geology of the area is described in deta il & 0stmo 1990) and hydrochemistry (Jerqensen by Longva (1987). The Romerike aqu ifer con­ et al. 1991). The aquifer is entirely fed by re­ sists of a 105 km' expan se of Quaternary ice- charge from prec ipitation. 0stmo's (1976) map NGU• BULL.422,1992 Oil-leakages on the 0vre Romerike aquifer 69

, Northern Romerike l , Hydrogeofogl:al jra'Clage maa , ,,0,"/ 20

-~ _ Areas jI.:JOVll marine I,mi\ ------11)(1

~~ ~------205 --

'''-208----

VERTlCALSAI1Pl.INGHOlES

15· ANGLED S"MPW'G 1l0LES N 140nrnSCAVENGER wELLS 100mm SCAVENGER ....ELLS

....-----.. PROFllESSHO ....NINFIGS.4&S I r OIL LEAKAGEOC(lIRfD BYP82 t

Fig. 2. Regional groundwater flow pattern within the 0vre­ Fig. 3. Summary map of leakage area at Trandum. Profiles Romerike aquifer - after 0stmo (1976), Jerqensen & 0stmo I-I' and 11-11' refer to Figs.4 & 5. The leakage occurred by PB2. (1990).

(Fig.2) indicates that the major central part of lZlvre-Romerike aquifer, ea. 1-2 km SW of the aquifer drains towards Hersjoen and the Hersjoen. Groundwater flow drains towards River Risa. The average discharge in the Ri­ Hersjoen. On October 12th, 1990, an acute ver Risa (in the period 1967-74) was 0.85 leakage of light fuel-oil (consisting mainly of m'ls (Jorgensen & lZlstmo 1990). The marginal C'" - C,,, hydrocarbons, and with a density of parts of the aquifer drain outwards towards 845 kglm' at 15°C) was discovered at an under­ springs in the periphery of the delta. Hydrau­ ground storage tank adjacent to Building 111 lic and hydrochemical balances for the aquifer at Trandum (Fig.3). The entire contents of the have been calculated. The permeability within tank (20,000 I) had drained rapidly out into the the aquifer is believed to range between 5 ground. On excavation of the tank, the leaka­ mId for coarse sands to 0.06 mId for silts on ge was found to have occured through a disc­ the basis of particle-size distributions (Jorgen­ rete hole of c. 2 cm diameter. It is believed sen & lZlstmo 1990). that the hole was caused by a sharp stone puncturing the (possibly corrosion-weakened) tank under loading. The incident was reported immediately and the Geological Survey og The Trandum oil-leakage Norway (NGU) were requested to undertake further investigations and immediate remedial Introduction action. The results of these investigations are The location of the Trandum base is shown reported in detail by Storro (1991). on Fig.1. It lies on the central part of the 70 Gaute Stor'" & David Banks NGU - BULL 422, 1992

Fig_ 4. Profile I - I' (Fig_3) through A PB~ IBuilding 111 I lB2 leakage area at Trandum, sho­ D wing THC content of sediment and' gas-chromatogram type. Drilling dates: P81 = 17·20/10/ 1\ 90; P82 = 23-26/10/90; 3 = 'i'I'\!\ 5 /i 27110/90; 8 = 7/11/90; P84 & D e e) \ 0 Gravelly sand / = 8/11/90. Assessment of «core coarse sand volume» is based on situation e ~.lle 10 as determined on November e / •• le th e /. 8 , 1990. e / • 15-~ _ \,\~ 15 e / • e I I' ~ • e () le 20 Ita / e I/ e - c • ~()/ " () /' e/ Fine sand { sill 25 e /~ I \ e,,/ ~ : I. GW I e ..... ----/.--" e ---. ----r------e ------~------= ------~- 30 I . e I . i ~ ~ 35 I () e e ~ , 40 KEY [) ,. 5000 mg/ kg hydrocarbon _ GC Iypel o 1000 - 5000 mg / kg -=:J Type 2 o 100-1000 mg{kg ~ Type 3 o <100 mg/kg -=:J Type 4 -- Core Yolume >1000 mg/kg ~ Indeterminate

Fig. 5_ Profile IJ - 11" (Fig.3) PB6 IBUildin g 6 PB2D 2 C 4 through leakage area at Tran­ 1111 dum, showing THC content of /,,- sediment and gas-chromato­ I gram type. Drilling dates: P82 = 23-26/10/90; 2 = 26/10/90; 1 , / I' -. 5 C, D & 6 = 8/11/90; P86 = Gravelly sand / /1 _) e 9/11/90. Assessment of «core I volume» is based on situation coarse sand / • {e !!'I 10 as determined on November 1 • I e !!'I 1 IJ I 8th, 1990. '11 ,e e o _ 15 / ~ _ le e e ~e -~e------IIf) ~ ()I/ ~",------j 20

I \ ~ { e !!'I !!'I Fine sand/sill '.... ~....() ';! _ I --__- ~ e co 25 ~ ~., ~ ______e "T.7GW ~e e- - __------30 , I ., () I e i !!'I !!'I !!'I 35 o (M) !!'I

~ o >5000 mg/kg hydrocarbon _ GC Iype 1 o 1000 - 5000 mg/kg IK:::J Type 2 o 100 - 1000 mg /kg = Type 3 o <100 mg/kg .-::J Type 4 -- Core volume >1000 mg!kg ea, 8.11.90. NGU• BUll. 422, 1992 Oil-leakages on the 0vre Romerike aquifer 71

Investigations I A plan of the boreholes constructed to investi­ I gate the leakage is shown in Fig. 3. The sca­ /~ venger borehole PB1 was established imme­

diately after the leakage was reported. Sedi­ r-----_ / -- - _10 0- ment and water samples were taken from PB1 <, - / to establish the extent of contamination. It was planned to pump the borehole continuous­ ly to scavenge contaminated water from the aquifer, but in the event, neither seriously ----- contaminated sediment nor contaminated wa­ ter was discovered in PB1. )j A new scavenger borehole PB2 was thus established on the site of the oil tank, and ~.\\ several trial holes were drilled in the surroun­ \.. \+ -, ding area. In these, oil contamination was found in the unsaturated zone down to c. 25 m un­ der ground level. The mineral oil content (total hydrocarbon content, THC) in the sediments N was determined by gas-chromatography, and the results of the analyses are shown in Figs. 4 & 5. Parallel with sediment-sampling, obser­ vation boreholes were established for monito­ t ring of groundwater level and hydraulic gradi­ ent. 10 20 30 40m

Stratigraphy The sediments sampled in the various investi­ Fig. 6. Groundwater contours (non-pumping) at leakage 'gation boreholes gave the following stratigrap­ area, Trandum, 1st November 1990. Dashed lines show topographical contours. Contours are in m above sea level. hy:

0- 19 m : layered sedirnents, comprising ma­ in groundwater level during the thaw period inly gravelly sand with individual is normally less than 1 m. pure sand or gravel layers. A pumping test performed on PB2 yielded 19 - 31 m: fine sandlsilty sand with gravelly a transmissivity coefficient of c. 3 m'ld, and a layers. hydraulic conductivity of around 0.4 mid (in 31 -? m: finesand, silty sand, silt. the coarser aquifer at Sessvoltmoen, the trans­ missivity was calculated as c. 100 rnvd). Using It is worth noting here that the stratigraphy a water table gradient of 0.012 - 0.018, and at Sessvollmoen was the opposite of that at an assumed effective porosity of 10 - 20 %, Trandum, i.e. fine-grained sediments in the one obtains a natural average groundwater unsaturated zone, with coarser sediments velocity of some 0.04 - 0.08 mid. Jerqensen beneath the water table. & l2Jstmo (1990) have reported velocities of up to 0.14 mid in the l2Jvre Romerike aquifer. The downward percolation velocity of Hydrogeology groundwater recharge in the unsaturated zone The water table in the Trandum leakage-area, has not been investigated during this project. in November 1990, was found to be 31-32 m Jerqensen & l2Jstmo (1990), however, report below ground surface. Contours on the water velocities of up to 0.2 mid during snow-melting. table (under non-pumping conditions) are shown in Fig. 6. The maximum water table level in the lZJvre Romerike aquifer normally Oil contamination within the sediments occurs around the beginning of July, at the Borehole 3 was drilled on October 27th , 15 end of the snow-melting period. The increase days after the leakage occurred. Significant 72 GauteStor", & David Banks NGU·BULL422,l992

rRISTA~E

GC TYPE I GC TYPE 2

Fig. 7. Qualitative assessment of mineral oil in sediment samples. The figure shows the four main types of gas-chromato­ gram observed during investigations at Sessvollmoen & Trandum. See text and Figs. 4, 5, 12 & 13 for further explanation. oil contamination was discovered down to 23 calculated to be around 940 m'. With a leaka­ m depth in the sediments in borehole 3. Thus, ge of 20 m' oil, this represents an average the initial phase of downward movement of concentration of 21 I oil/m' sediment, or 10.4 the oil-body during the leakage from the Tran­ g/kg (dry weight) assuming a dry density of dum oil-tank appears to have occurred at 1700 kg/m'. This agrees well with figures gi­ around 1.5 m/day. Borehole PB2 was drilled ven in the literature (15-25 l/rn') for oil-retenti­ on October 26th , and borehole D was drilled on capacity in this type of sediment (Kristian­ on November 8th at the same location. The sen 1983, Snekkerbakken 1988). 'core' of oil contamination extended to over The core volume does not appear to have 24 m deep in borehole D, as opposed to 15 encountered the water table by November 8th • m deep in PB2. Thus, in this later phase, the However, the calculated downward percolation vertical seepage velocity was at least 0.7 m/ velocities indicate that, although a major pro­ day. portion of the oil has been retained in the The assessed extent of the oil contamination unsaturated zone, the core volume is likely to in the unsaturated zone (as of November 8th) have reached the water table shortly after is shown in Figs.4 & 5. It has been. chosen that date. The lack of recovered oil phase to define the contamination's 'core volume' (see next section) in the scavenger wells indica­ as that volume of the sediments exhibiting THe tes, however, either that any floating oil-phase values in excess of 1000 mg/kg (the Dutch on top of the water table is of very limited B-value for soils). The core-volume appears extent, or that the oil is completely dissolving to extend in a coniform manner from the tank in the saturated zone's groundwater. area (PB2), and under Building 111 in a wester­ Gas-chromatography is a useful tool for the ly and southerly direction. The core-volume is qualitative assessment of the hydrocarbon NGU · BULL. 422.1992 Oil-leakages on the 0vre Romerike aquifer 73

content in var ious parts of the contaminated the water scavenged from the site. The lack area . Fig.7 sho ws the four main types of chro­ of oily taste or sme ll may conceivab ly be due matogram resu lting from the analyses. The to the lighter, taste-causing, oil components prominent peaks in the chromato grams repre­ being evaporated or biodegraded during th eir sent the n-alkane components with incr easing passage through the unsaturated zone (Storm numbers of carbon atom s in the chain. The 1991). It should be noted that samples taken chromatograms for the samples with the hig­ from the scavenger well in January 1992, and hest oil con centrations (> 5000 mg/kg) show ana lysed by another laboratory, appear to indi­ fresh oil (GC typ e 1), with a GC maximum cate a significant reduction in the hydroc arbo n around C". Sampl es with an oil content 1000 content of the groundwater one yea r after - 5000 mg/kg yield chromatograms whe re the NGU's investigation (SIFF 1992), probably in maximum is displaced a little to the righ t. typi ­ part due to bioremediation attempts. cally up to C,_, due to evaporation of lighter Reference samples were taken from a supp­ components as the oil seeps down into the ly borehole at the Trandum military base some unsaturated zone (GC type 2). GC type 3 is 800 m NNE of the tank-leakage (Fig. 1). The typ ical of the 'natural background' signal occur­ groundwater from this borehole contained cle­ ring in uncontaminated so il due to naturally ar traces of mineral oil. Subsequent samples occurring hydrocarbons. Here the chromato­ have been take n for deta iled analysis. These gram max imum is broa d and occurs far to the analyses sho w partially dissolved organ ic car­ right, typ ically arou nd Coo - C" . Components bon in concentrations higher than the Nor­ of GC typ e 3 cou ld be observed in les s conta­ weg ian drink ing water standards permit (TOC min ated samples « 1000 mg /kg ). GC type 4 < 3 mgll - SIFF 1987) . The geometry of the is characteristic of oil strongly enrich ed in situation appears to preclude the oil-tank leaka­ heavi er components, due to extens ive evapora­ ge as a possible source. It is therefore sugge­ tion or biodegradation of lighter components sted that there may exist other sources of oil (biodegradation may , in add ition to the end contaminat ion to groundwater in the area. It products CO, and water, also produce heavy , should, however, be noted that a later samp­ long-chain by-products, thu s emphas ising the ling round (SIFF 1992), early in 1992, failed right-hand end of the chromatogram even to detected significant hyd roca rbo n contamina­ more (Stale Jo hnse n pers.comm. 1991)). The tion in the supp ly borehole. GC type 4 maximum may lie as far to the right The inorganic chemical composition of 0vre as C,-. GC type 4 is typical of heavily evapora­ Rom erike's groundwater is documented by ted/degraded fuel oil or of relatively fres h lubri­ Jerqensen et al. (1991), and analyses from the catio n oil (i.e. possible co ntamination dur ing Trandum military base show good agreement drill ing). The ch romat og ram types observed in with the res ults in thei r paper . It should howe­ the samples from Trandum are shown in ver be noted that the gr oun dwater at Building Figs.4 & 5. 111 co ntains sign ificantly higher concentrati­ ons of nitrate (2 mg NOJ'-N /I) than that which is typical for the aquifer. This nitrate co uld be derived from oil decomposition, or from Oil contamination of groundwater other local pollution sources such as Tr andum Groundwater samp les were analysed for both landfill (Misund & Seether 1991, Seether et al. tota l hydrocarbon content (THC) by gas-chro­ 1992). matography, and total orga nic carbon (TOC). The resu lts indicate a clear increa se in the groundwater THC from the fir st sampling date Remediation strategy (7/11/90) to the more rec ent samples (Fig.14). Scavenger we ll PB2 is operated continuously By 20/12/90 all the scavenger wells (PB 1 ­ to prevent further spreading of oil contaminati­ PB6) exhibited THC in the range 0.6 to 1.6 on in the saturated zone away from the site. mgll, i.e. higher than the Dutch C-value. It is Scavenger wells PB1 & PB3-6 are equipped thus co ncluded that the incident at Trandum with submersible pumps, and can be brought base led to loc al cont amin ation of groundwa­ into operation immediately sho uld the situation ter in the immediate vicinity of the leakage, deteriorate. although no distinct oil-phase, nor any taste or sme ll of oil, has hith erto bee n observed in 74 Gaute Storm & David Banks NGU • BULL. 422. 1992

SESSVOLLMOEN

EIUIlO lJ

• \I .. ll.8(_(l.. o r t i,P.l(.. u ~ .,

10 20 30 m ======\; I

Fig. 8. Summary map of leakage area at Sessvollmoen. Profiles 1·1' and 11-11 ' refer to Figs. 12 & 13,

The interested parties in the tank-leakage inci­ ge occurred near the entry pipe to the return dent are agreed on a strategy of in-situ bio­ tank. and was due to excavation work in the remediation for the contaminated area. The vicinity. The leakage was reported to the pollu­ remediation has been planned by the Nor­ tion authorities. and GU was again invited wegian Geotechnical Institute (Hauge 1990). to investigate the leakage. A scavenger well and a pilot project is curr ently underway (Bre­ was sunk to 24 m at point PB (Fig.8). about 6 edveld et al. 1991). m from the assumed point of leakage. The well was completed on December 19t • 1990. The scavenger well was continuously pumped The Sessvollmoen oil-leakage at 1800 IIhr, such that pump sat in the pum­ ping water level at a depth of 19,65 m in order to recover any oil phase floating on the water Introduction table. As a result of the Trandum leakage. aware­ ness of potential leakage prob lems was un­ doubtedly increased within the military authori­ Investigations ties. As a result, it was not long before a These consisted of: second suspected leak was reported, at Sess­ (a) drilling sampling boreholes and piezome­ vollmoen base (Fig.1), which lies a little to the ters to investigate stratigrap hy and hydrogeo lo­ NW of the central part of the aquifer . Accor­ gy ding to 0 stmo's (1976) map. groundwater drai- (b) sampling the sediments in these boreholes . nage in this area is ESE tow ards the River and analysing them for hydrocarbon content. Risa. On November 2nd , 1990. the under­ (c) taking regular water samples from the sca­ ground fuel tanks adjacent to Building 14 (Fig.8) venger well for THe , Toe and inorganic analy­ were routinely inspected. without any indication ses of a leakage. Reinspection occurred on Decem­ ber 13th , 1990, and calculations of consumpti­ on revealed that a volum e of up to 10,000 I Stratigraphy of light fuel oil appeared to have been lost Sediment samples were taken in the pumping thro ugh leakage.It is assumed that the leaka- borehole (down to 24 m) and several investiga- NGU • BUL L. 422, 1992 Oil-leakages on the 0 vre Romerike aquifer 75 tion boreholes (down to 20 m). The sediments encountered were mostly fine sand and silt above the water table. In the pumping boreho­ le, where sampling intervals were more frequ­ ent than in the other holes, a c. 4 m thick layer of relatively well-sorted medium sand was encountered at around 8 m depth . A similar, thinner layer was also found at "around 15 m depth . The water table was encountered at c. 17 m. Below the water table , fine sand and I silt continued down to c. 20 m, below which coarser sediments (poorly sorted medium­ ~ grained sands) were found again. The base of this coarse layer was not encountered in Fig. 9. Estimated groundwater contours (non-pumping) at the 24 m deep pumping hole. The aquifer thus leakage area. Sessvollmoen, 11th April 1991. Conto urs are in m above sea level. appears to be a layered aquifer, with alterna­ ting higher and lower permeab ility layers . Verti­ cal flow between layers in response to pum­ ping is likely to be important. It will be noted that the situation at Sessvoll­ moen is strat igraphically the oppos ite of the situation at Trandum, where coarse grained sediments in the unsaturated zone overl ie fi­ ner sediments below the water table.

Hydrogeology The pumping well was installed to 24 m (with slotted screen from 12 - 24 m) and is thus believed to have drawn its water largely from the coarse-grained layer. All ptezorneters we­ Fig. 10. Groundwater contours (pumRing from borehole PS) at leakage area. Sessvollmoen, 11t April 1991. Contours re, on the contrary, installed to 20 m (with the are in m above sea level. A·A · is the catchment area of exception of piezometer 11, installed towards the pumping borehole. the end of the project) and so measured condi­ tions near the base of the finer layer. the aquifer's rest condition. However , by obser- The rapid establ ishment of the scavenger I ving the behaviour of piezometers during the well and the requirement to keep it continuous­ recovery tests , it was possible to estimate the ly pumping permitted only limited hydraulic rest condition with a fair degree of accuracy testing in the form of recovery tests of up to (Fig.9). The natural groundwater gradient was one hour's durat ion. The conf iguration of the found to be towards SSE rather than ESE as tests was such that many of the assumptions indicated by 0stmo (1976). It was thus poss ib­ of common aquifer test methods were not le to calculate the drawdown caused by pum­ fulfilled, allowing only an approx imate estimate ping. It appeared that only two of the observa­ of the aquifer's apparent transm issivity (T). tion boreholes experienced a drawdown of Using the Jacob approx imation on the pum­ greater than 2 cm during pumping (compare ping well, and piezometers 1 and 6, and also Figs.9 & 10), namely piezometers 6 (5 m from by applying the Theim method (Krusemann & pumping well, drawdown 5 cm) and 1 (1 m DeRidder 1989), T was estimated as c.100 from pumping well, drawdown c.30 cm). From m' /d. Assuming an aquifer thickness of seve­ this information one can const ruct the scaven­ ral tens of metres, this implies an average ger well' s catchment area either empirically permeability of a few mId , agreeing well with by superimpos ing the cone of drawdown upon Jerqensen & 0 stmo's (1 990) estimates. the natur al groundwater gradient (Fig.11) or Due to the immediate operation of the sca­ analytically by using the followi ng formulae venger well it was not possible to docu ment (Mutch 1989): 76 Gaute Storm & David Banks NGU · BULL. 422. 1992

r- --:-:=-=~'"=:_::_---HI O ] 1 --- I -s" lm I ~ ..- -, , '80 , , ./ 36 " " 180 35 - / " \ / / -10 - --<, \ I - / 1803l. _ \ / " \ I / / "\ I 'BC _ \ 1 n " /"-- -': 30 \ I .I I .', \ I \ WE LL I I

\\ " <, _ ...... " / I 180 31 _ " \ / I " / 180 30 / " .... »<..- I I 18° 29 ~

",,-,,1I!lOze _

Ic------=-=-- -=O"::""---'''' 17

- '''' "

.", - ., / " ., I --- ...... " \ I / - "" " I \ / ' J I <, " \ _ 1 ~ JS I -, - \ f I / \ I ( ~~\ I I I I I - \ .... =. / 30 \ / - T :-'80 ,/ \ \ I I \ / \ " J \ <, / - "'" I!I " ...... / I \ _-- I - 10 lO - " -, ...... - " '"" .... - - - -- ~ - '''' " ,,-- 18" - " I CO NTOURS ON REST WATE R L EVEL CO NTOURS ON PUMPING WATER LEV EL (m ab ove sea level ) (cm + l e Orn above sea level) CO NTOURS ON DRAWD OWN (c m) FLOW STREA MLI NES - WELL 'S CATCH MENT AREA

Fig. 11. Empirical calculation of scavenger borehole's catchment area , by super impos ing cone of depression onto the natural groundwater gradient, See text for further explanation.

L = Q There was found to be good agreement betwe­ en the two methods, althoug h the analytical ri method obv ious ly depends on having a reliab­ le value of transmissivity. The empirical met­ .and rps L hod gave a catch ment area as shown on Fig. 2· 7t 10 with L = c. 25 m and rps = c. 5 m. The Where L = maxi mum breadth of catchment existing catch ment area cannot be regarded area upgradient of pumping well as totally satisfactory, particularly when one ros = extent of catch ment area down­ cons iders that an oil body will spread horizon­ gradient of pumping well tally during its passage through the unsatura­ T = transm issivity ted zone. The boiler house, pipes and part i =natural groundwater gradient of the tanks lie outs ide the catchment area . and Q = pumping rate. Unfortunately, the deep piezometer, no.11, NGU - BULL. 422. 1992 Oil-leakages on the f2Jvre Romerikeaquifer 77

Fig. 12. Profile I - I' (Fig.B) through leakage area at Sessvollmoen. showing THC content of sediment Building 14 and gas-chromatogram type. Drilling dates: PB ., 8 7 PBl 9 5 17-19/12/90. boreholes L-----_----l. --_------__-----rO 5·10., 19-21/3/91.

" ~ ~ Fine sand/silt 3 /C_IJ ....." £ 6 \ <, ,\ 6 l,----~~-~"...... -11-}- '" o ~ . \,IS/~-- 9 '" 'Ii--- e " 12 I -() Fine sand I ~aendJum & t> 15 GV t:, £ () _ _. _.. -t.. ····I7::":"_·····U·······_() _ . ; () ~ Fine sand 18 i () -_£_- _._()-_. -- 21 t> Poorly sorted medium sand o 24 (M)

KEY o > 5000 mg/kg hydrocarbon _ GCtypel o 1000 - 5000 mg/kg -=:J Type 2 o 100 • 1000 mg/kg ~ Type 3 o <100 mg/kg ~ Type 4 6 <10 mg/kg Indeterminate -- Core volume >5000 mg/kg -- Core volume >1000 mg/kg

Fig. 13. Profile 11 - II' (fig.B) through leakage area at Sessvollmoen, showing 10 PBl 6 THC content of sediment ------,----,--,------.0 and gas-chromatogram type. Drilling dates: PB ., /? 3 17-19/12/90, boreholes t 1 Silt I finesand 5-10., 19-21/3/91. 11 ...... 1\ ,)/ 1 6 -- ~-~"-f---- 9 \'=t=j'''/I __ . __Medium sand Silt I finesand 12 ---+--- Medium sand ---t--- - .- 15 GV ------~--- ._ _ ·1············_··················Silt I finesand 18

21 Poorly sorted medium sand 24 (M)

KEY IT> 5000 mg I kg hydrocarbon _ GC type 1 01000-5000 mg/kg -=:J Type 2 o 100-1000 mg/kg ~ Type 3 o 10-100 mg/kg w::::J Type 4 c: <10 mg/kg ~ Indeterminate --- Core volume >5000 mg/kg -- - Core volume> 1000 mg I kg 78 GauteStorre & David Banks NGU-BULL422.1992

-was not installed when the recovery testing were found; c. 80 mg/kg down to c. 20 m, was carried out. It was later shown, however, and lower values of 23-76 mg/kg' down to 24 to respond much faster, and to a greater ex­ m confirming the earlier assumption that the tent, than piezometer 1 to changes in pumping majority of downwashed oil remained in the rate, indicating a considerably higher permeabi­ upper, finer layer of the saturated aquifer. lity in sediments below 20 m than in those One spurious value was obtained, 2400 above. Vertical flows between the aquifer lay­ mg/kg at 8m deep in borehole 5. This, by the ers may thus be important, but within the sco­ situation's geometry, is unlikely to be due to pe of the project it was not possible to investi­ the leakage under investigation. Although it is gate these further. As any oil contamination not believed to have been a major problem which reaches the water table is likely to be with the sampling programme, contamination largely confined to the vicinity of the water during drilling could conceivably have led to table (Le. to the finer layer) due to its lower this anomalously high value. It could also be density relative to water, the calculation of the due to pollution from another source. scavenger well's catchment area on the basis In summary, the core of highly contaminated of piezometers in the finer layer is defensible sediment is relatively limited, being located in for practical purposes. the upper part of the unsaturated zone (Figs. 12 & 13). As it was not possible to drill boreho­ les in the exact vicinity of the suspected leaka­ Oilcontamination within the sediments ge, the core volume's area! extent to the west Sediment samples were extracted from the is to some degree unknown. By assuming, pumping hole and from investigation holes 5, however, that the core volume has a double 6, 7, 8, 9 and 10 using a suitable throughflow cone shape, it has been possible to estimate sampling device. The holes were bored by the that c. 350 m' of sediment are contaminated ODEX method using only air as a drilling flu­ to over 5000 mg/kg, and c.600 m' to over id, and soya oil where lubrication of the bit 1000 mg/kg. Assuming that the leakage was was necessary (this does not affect analysis a maximum of 10 m' of oil, the mean oil con­ results for mineral oils). The samples were tent in the c.600 m' of sediment would be analysed for Total Hydrocarbon Content (THC) 16.6 l/m' or 8 g/kg dry weight. With an assu­ by gas chromatography. The results of the med porosity of 0.36, this is equivalent to 4.6 THC analyses are presented in Figs. 12 & 13. % saturation with oil. As has been described above, the chromato­ The concept of a retention capacity (Le. that gram types were identified, and these are also a sediment will have a capacity to retain a indicated on Figs. 12 & 13. certain content of oil in relatively immobile As can be seen from Figs.12 & 13, serious form due to adsorption, capillary retention etc) oil contamination was only found in the scaven­ has been widely used. CONCAWE (1976) and ger well (PB) and investigation hole 7. The Testa & Paczkowski (1989) give the retention sample from 3 m in the scavenger well yielded capacity for fuel oil in unsaturated finesand no trace of mineral oil contamination, while the and silt as between 40 (kerosene) and 80 sample at 4.5 m contained c.14 g/kg'. This (more viscous oils) IIm'. Testa & Paczkowski indicates that oil has spread rapidly horizontal­ also relate that the API uses a retention capaci­ ly from the leakage point. During drilling a ty of 15% of available porespace. The figures 5-10 cm thick clayey layer was encountered deduced for Sessvollmoen are about one third at 4.5 m, and it seems likely that the oil has of these values. This indicates that the majori­ spread extensively on top of and within that ty of the sediment within the core-volume layer. Within the sediments recovered from the contains a content of oil lower than the theore­ pumping well, the oil content had declined to tical retention capacity. The volume of sedi­ under 5000 mg/kg by c.7 m deep, to under ment saturated with oil to retention capacity 1OOOmg/kg by c.11 m and to under 100 mgl kg by c. 13 m deep. Even under 13 m deep the oil concentration appears to decline steadi­ 'Trials using sediment samples with a known content of hydrocarbon indicated that it was only possible to extract ly with depth down.to the water table (23 (using rnethanol/n-pentane and dichloromethane) c. 90% mg/kg), indicating a small but finite down­ of THC for analysis (John sen, 1991). Measured values are quoted in the text, but in Figs. 4. 5. 12 & 13 sediment washing of oil components to the water table. analysis results have been multiplied by 1.11 to give a «cor­ Below the water table slightly higher values rected" THC. NGU• BULL 422,1992 Oil-leakages on the((Jvre Romerikeaquifer 79 is thus probably limited to a relatively small mg/I volume immediately under the leakage area. 100 The fact that the average oil content of the core-volume appears to be considerably less than the sediment's theoretical retention capa­ • PS SESSVOLl city bodes well for the potential pollution thre­ at from the site. There would appear to be little danger of further substantial downward movement of the oil phase through the sedi­ ment.

Oil contamination of groundwater In the water recovered from the scavenger borehole, there has been observed neither smell, taste nor film of oil. Gas chromatograp­ hy has also indicated that the oil found in the / water from the borehole consists only of solub­ I Legend 0,1 6 • re le, and not oil-phase, components. This is )l TOe (acidified/aerated) o THe A. B, C. Dutch limits, THe consistent with the majority of the oil-phase SIFF.Norwegian limits, TOe contamination being retained in the unsatura­ ted zone, with a limited degree of down­ washing of water-soluble oil components in percolating recharge-water. The THe and TOe 0,Q11+------,-----.--.-----,-----,---- 30/1 contents of the abstracted water from the 1/11 1/12 31/12 1/3 31/3 1/5 1990 I 1991 scavenger borehole are shown in Fig.14. A very high oil content of 2.4 mg/l was observed Fig, 14. Variation of THC and TOC in groundwater from in January 1991. This could conceivably be Trandum and Sessvoltmoen with time. «Water supply bore­ due to contamination of the sample with oily hole» = Supply borehole for Trandurn military base, situa­ ted at NW end of lake Transjeen (Fig. 1).• = TC = total material from the unsaturated zone during carbon (measured prior to acidification and aeration of boring. Since January, the oil content of the sample), X = TOC = measured after acidification and aerati­ water has declined steadily with time, until by on (total organic carbon, less some of the more volatile hydrocarbon components), 0 = THC = total hydrocarbon April it was 0.045 mg/l, only a little over the content. aSH 6 = piezometer 6, Sessvoltmoen. Dutch A-value (0.02 mg/l). The decline in oil content with time is thought to be due to the progressive drawing in of less contaminated groundwater from the NW, and from deeper procedures such as active biorestoration. Pro­ levels in the aquifer (the borehole only being vided that the scavenger well is relocated partially penetrating) with time. A water samp­ such that its catchment area encompasses the le taken from peizometer 6 on April 15th, 1991 entire contaminated area, there will be little had a rather high THe value of 0.62 mg/l. danger of contaminated groundwater sprea­ Penetrating only just below the water table, ding from the site. Passive rehabilitation of the having bee" pumped for only a short time aquifer (Le. allowing the oil to naturally be prior to sampling, and lying outside the scaven­ broken down in-situ by microorganisms) is ger borehole's catchment area, the sample thought to be satisfactory. probably represents the content of hydrocar­ bons concentrated near the top of the satura­ ted zone in a relatively undisturbed part of the Conclusions aquifer. Two leakages of, respectively 20 m' and <10 The lack of existing groundwater abstracti­ m' of fuel oil have been investigated on Nor­ ons in the vicinity of the leakage, the limited way's largest discrete aquifer, the Q.lvre Rome­ size of the incident and the apparent capacity rike aquifer. The first leakage, of 20 m', occur­ of the unsaturated sediment to retain the red where the unsaturated zone consists large­ majority of the oil contamination argue against ly of coarse sand, with finer sediments below the establishment of costly aquifer remediation the water table. Despite an average oil retenti- 80 Gaute Storm & David Banks NGU - BULL 422. 1992 on of 21 I/m' in the unsaturated zone, the oil the degree of groundwater contamination is appears to have migrated downwards relative­ minor. The contrast between the two case ly fast (up to 1.5 m/day) during the leakage's studies illustrates the importance of the type initial phase. Hitherto, no distinct oil phase of sediment in the unsaturated zone, and thus has been observed in the installed scavenger its retention capacity, when assessing the boreholes. Dissolved mineral oil concentrations impact of oil spills on aquifers. of around 0.6 - 1.6 mg/I have been observed in the pumped water, and these showed no Acknowledgement signs of decreasing with time during the peri­ Thanks to Tidemann Klemetsrud. Helge Skarphagen. Arve Misund and NGU's drilling crew for performing most of the od of NGU's study (although subsequent mea­ field work involved in these studies. to Stale Johnsen (for­ surements in early 1992 do indicate some merly of IKU. now with Statoil) for useful discussion of hydrocarbon analyses. to the canteen at Trandum for some decline). jolly fine breakfasts and to Bjern Bmnstad of the Nor­ In contrast, the second leakage « 10 m') wegian Defence Construction Service for allowing us to occurred where the unsaturated zone consists publish these data. largely of fine sand and silt, with coarser, poorly sorted medium sands under the water table. In this case the oil body is retained at a relatively high level in the unsaturated zone, with only very small « 100 mg/kg) concentrati­ ons of hydrocarbon being observed lower in the unsaturated zone. Despite initially high (2.4 mg/I) hydrocarbon concentrations in the scavenged groundwater, these have substanti­ any decreased during the study period to 0.045 mg/1. This is consistent with the bulk of oil being retained in the unsaturated zone, with only minimal downwashing of dissolved hydro­ carbons in recharge water. Both studies are open to criticism on seve­ ral grounds. It would have been desirable to drill a more extensive sampling network at both sites. This was not possible due to both economic and logistical constraints (e.g. avoi­ dance of cables, pipes and underground tanks). The latter precluded intensive sampling of the area under Building 111 at Trandum, and the area immediately around the tanks at Sessvoll­ moen. It would also have been highly desirab­ le to take further time-series of sediment samp­ les, to monitor the downward seepage of oil at Trandum, and to verify the retention of the oil's core volume at Sessvollmoen. Sampling and analysis procedures appear to have been reliable, particularly at Sessvollmoen where there was little indication of contamination during or after sampling, and where a consis­ tent picture of decreasing oil concentations through the unsaturated zone down to < 100 mg/kg was obtained. Neither oil leakage produced consequences remotely approaching the «doomsday scenari­ os» predicted by the media. In both cases, groundwater contamination appears to be lo­ cal and controllable by appropriately sited scavenger wells. In the case of Sessvollmoen, NGU - BULL. 422, 1992 Oil-leakages on thefl!Jvre Romerikeaquifer 81

References Misund, A. & Saether, O.M. 1991: undersekelse av foruren­ Banks. D. 1991: Kartlegging av 0ljeforurense1 grunn/grunn­ set grunn og grunnvann ved Trandurn militaerleir. Nor. vann ved bygning 14, Sessvollmoen militaerleir. Nor. geol. unders. report 91.228, 137 pp. geol.unders. report 91.190. 148 pp. Morland, G., Folkestad, B., Hauge, A., Kolstad, P., & Fors­ Breedveld. G.D., Kolstad, P., Hauge. A., Briseid, T. & Bren­ berg, C.F. 1990: Kartlegging av spesialavfall i deponier stad, B. 1991: In situ bioremediation of oil POllution in og forurenset grunn i Akersnus fylke. Nor. geol. un­ the unsaturated zone. NATO/CCMS, Pilot study, 5th ders. report 90.084. 170 pp. International Conference, Washington, 18-22Nov. 1991. Mutch, R.D. 1989: Aquifer restoration: basic principles. CONCAWE 1979: Protection ofgroundwater from oil polluti­ Proc. course - «Clean-up of waste-disposal sites», 13-15 on - CONCAWE's Spesial Task Force No. 11, CONCA­ novo 1989, Centre for Professional Enhancement, Den WE, Den Haag, April 1979. Haag. Davidsen, F. 1990: Gardermoen og vannforurensende aktivi­ Norwegian National Committee for the IHD 1973. Hydrologi­ teter. Vann, 3, 216-225. cal data - Norden. Representative basins; Romerike, Falkenmark, M. (ed.) 1972: Hydrological data - Norden. Norway. Data 1965-71. Norwegian National Committee Representative basins, introductory volume. National for the International Hydrological Decade. Oslo, 1973. Committees for the International Hydrological Decade Norwegian National Committee for the IHD 1975. Hydrologi­ in Denmark, Finland, Iceland, Norway & Sweden. cal data - Norden. Representative basins; Romerike, Hauge, A. 1990: Plan for rensing av oljeforurenset grunn Norway. Data 1972-74. Norwegian National Committee pA Trandum. N.G.I. (Norges geotekniske institutt) re­ for the International Hydrological Decade, Oslo, 1975. port 902542-01. SIFF 1987: Kvalitetsnormer for drikkevann. SIFF (Statens Kristiansen, H. 1983: Virkning av oljeforurensning i grun­ Institutt for Folkehelse) guide «Drikkevann G2». nen. NIVA (Norsk Institutt for Vannforskning) report SIFF 1992: Trandum - analyse av vannprever. Letter from 0-82112. Statens Institutt for Folkehelse to Forsvarets bygnings­ Johnsen, S. 1991: Analyse av hydrokarboner i jord. IKU ­ tjeneste, Hamar, 24th February, 1992. SINTEF group report no. 22.1885.34/0//91. Snekkerbakken, A.M. 1988: Forprosjekt om sikring av grunn­ Johnsen. S. 1991, pers.comm.: StAle Johnsen, IKU (Nor­ vannsressursene pA Gardermosletta. VIAK AB report, wegian Continental Shelf and Petroleum Technology 15th October 1988, Karlstad. Sweden. 48pp. Research Institute). Trondheim, to David Banks, 1991. storre, G. 1991: Kartlegging av oljeforurenset grunn/grunn­ Jerqensen, P. & 0stmo, S.R. 1990: Hydrogeology in the vann ved bygning 111, Trandum militaerleir. Nor.geol. Romerike area, Southern Norway. Nor. geol. unders. unders. report 91.155. 71 pp. 418. 19-26. Saether, O.M.• Misund, A., 0degArd, M., Andreassen, B.Th. Jerqensen, P., Stuanes, A.O. & 0stm9. S.R. 1991: Aqueous & Voss, A. 1992. Groundwater contamination from a geochemistry of the Romerike area. Southern Norway. landfill at Trandum, Southeastern Norway. Nor.geol. Nor. geol. unaers. 420, 57-67.. unders. 422 (this volume). Kruseman. G.P. & De Ridder, N.A. 1989. Analysis and eva­ Testa, S.M. & Paczkowski, M.T. 1989: Volume determinati­ luation of pumping test data. 2nd edn. International on and recoverability of free hydrocarbon. Groundwater Institute for Land Reclamation and Impr.ovementllLRI Monitoring Review. Winter 1989. 120-128. . Publication 47. Wageriingen, Netherlands. 377 pp. Wolden, K. & Erichsen, E. 1990: Compilation of geological Longva. O. 1987: Ullensaker1915 11. Beskriverse til kvartaer­ data for use in local planning and administration. Engi­ geologisk kart M 1:50 000. Nor.geol.unders.Skr. 76. 39 neering geol., 29, 333-338. pp with map. 0stlandskonsult A/S, Geofuturum A/S. Akershus fylkeskom­ Misund, A., Banks, D., Morland, G:, Brunstad, H. 1991a: mune & Luftfartsverket 1991: Hovedflyplass Gardermo­ Norwegian national survey of hazardous waste in land­ en; grunnvannet som drikkevannskilde; en utredning fills and contaminated ground. Waste Management av grunnvannet som aktuelldrikkevannskilde for Romeri­ Today (News J.). 4, No.8, 30-35. ke. Misund, A., Morland, G., arunstao. H., Banks, D. 1991b: 0stmo, S. R. 1976: Hydrogeologisk kart over 0vre Romeri­ Kartlegging av spesialavfall i deponier og forurenset ke, M 1:20.000. Nor. geol. unders. grunn - Sluttrapport. 5tatens forurensningstilsyn report 91/01. 53 pp.

Manuscript received December 1991; revised typescript accepted May 1992.