ABSTRACT
The NOVA Chemicals Plant at Joffre, Alberta is undergoing an expansion of its production facilities. The expansion will require upgrading the site's water supply and treatment and wastewater handling systems.

The Red Deer River is the source of water used on site. The main demand for water is from the various cooling systems. Source water is reduced by cold lime softening prior to use. The spent lime from this process is currently land applied onto local farmland.

Waste streams generated on site include cooling tower blowdown which is relatively high in orthophosphates - up to 15 mg/L - that need to be reduced to less than 2.5 mg/L (1 mg/L as phosphorus) prior to discharge back to the Red Deer River.

A detailed bench scale and follow-up pilot scale program was undertaken to evaluate phosphorus removal by chemical coagulation using the spent lime from the water treatment process as a source of alkalinity and an aid to coagulation.

This paper will discuss the rational for looking at this and various other chemical combinations to address this problem, and will present the full results of the investigation. The impact of waste stream characteristics variability are discussed. Limited success was achieved at ambient pH. A high pH process, which relied on achieving optimal pH of 10.5 for precipitation of phosphorus was more successful and produced a settleable floc. In either case the addition of spent lime was beneficial to the process in reducing phosphorus levels and also reducing coagulant dose.

The success of this technology could see its potential application to municipal utilities that practise cold lime softening and need to address phosphorus removal at their wastewater plant.

KEYWORDS - Wastewater, Phosphorus, Chemical Coagulation, Clarification, Lime Sludge, Orthophosphates

INTRODUCTION
Preface
This paper presents the findings of a pilot and bench study that evaluated phosphate removal from cooling water blowdown. The study was conducted by Associated Engineering and Infilco Degremont Inc. (IDI) from April to August, 1998 at the NOVA Chemicals Ltd. plant located near Joffre, Alberta. The study evaluated the use of a

Degremont Densadeg clarifier for phosphate removal and established design criteria for a proposed phosphate removal plant.

The need for phosphate removal is found in Alberta Environmental Protection's (AEP) Standards and Guidelines for Waterworks, Wastewater and Storm Drainage Systems, 1997, which states that effluent discharged to water bodies must contain less than the maximum allowable concentration of 1.0 mg/L phosphorus (equivalent to 3.0 mg/L total phosphate). This is the most stringent phosphorus limit in the guidelines and applies to municipalities with population greater than 20,000. NOVA Chemicals Ltd. have committed to meeting this performance standard.

Background
The NOVA Chemicals Ltd. Joffre Plant is presently undergoing an expansion. The existing cooling towers, which serve E1 and E2, each have different chemical treatment programs. These programs include the addition of phosphates to reduce fouling and deposition of unwanted impurities in the cooling water process stream. The expansion includes the construction of several additional cooling towers which will increase effluent discharge along with phosphate loadings.

Scope of Study
The most important clarification design variable to be evaluated was the rise rate, as this would determine the overall size of the clarifiers. Additional objectives of the study were to:

• Verify, at a pilot plant scale, optimum chemical dosages determined in the bench scale study.
• Determine lowest achievable phosphate levels in cooling water blowdown.
• Evaluate both alum and ferric chloride as coagulants.
• Evaluate various types of polymers.
• Determine optimal spent lime concentrations (volumetric and gravimetric).
• Analyse solids concentrations throughout the process for design and operational guidelines.
• Analyse the effluent on site and at a certified laboratory to compare and verify results.
• Analyse sludge from the various processes for metals and nutrients.
• Determine need for filtration of phosphate effluent and optimal filtration rates and media types.
• Perform jar tests using similar processes utilized in the Densadeg clarifier.
• Perform jar tests simulating the addition of LAO, a waste product from the proposed AMOCO processing plant.

Note: Phosphorus levels in this paper are expressed as phosphate, where 1 mg/L of phosphorus is equivalent to 3 mg/L of phosphate (PO4). Total phosphate to orthophosphate ratio is discussed later.

EQUIPMENT AND METHODS
Pilot Plant Equipment
Figure 1 is a schematic of the pilot plant. The pilot plant consisted of a 22.5 m3/hr Densadeg clarifier followed by two filtration units. The clarifier consisted of a flash mixer/flocculator, reactor vessel and a thickener/clarifier vessel. Coagulant was injected at the flash mixer/flocculator, while spent lime, fresh lime and polymer was added at the reactor vessel. The reactor vessel also received recycled sludge from the thickener/clarifier vessel. Supernatant from the thickener/clarifier vessel flowed up through tube settlers to be collected in launders and gravity fed to two clear acrylic filters, Filters #1 and #2 on Figure 1. The clear filters allowed for visual inspection of the clarified effluent. Each filter contained the same type of media and configuration, as indicated in Table 1. The two filters were operated at different filtration rates.

Table 1. Filter Media Configuration

Filter Number	Description
Filter #1 & 2	300 mm of 0.5 mm effective size sand 300 mm of 1.0 mm effective size anthracite

Figure 1. Pilot Plant Flow Diagram (click image for more detailed figure)
The on-line analytical instrumentation used for monitoring water quality parameters consisted of a Hach turbidity meter, Wika head loss sensors, a Hach pH analyser, and a flow meter.

Coagulants And Polymers
The coagulants and polymers used were as follows:

Aluminum Sulphate 49% active and SG of 1.3	Reported as active ingredient mg/L
Ferric Chloride 49% active and SG of 1.34	Reported as active ingredient mg/L
Liquipam 452 long chained cationic polymer	Reported as neat
Jaschem A305 anionic polymer	Reported as neat
Nalco 8103 cationic polymer	Reported as neat
Betz 1115L cationic polymer	Reported as neat

Polymers were allowed to age for minimum of one hour before dosing. Maximum age of polymers used in the study was five days.

Aluminum sulphate and ferric chloride was dosed as mg/L active chemical where 100 mg/L (active) would equal 100/(0.49*1.34)=152 mg/L of neat chemical. All other chemicals were dosed as neat, also known as wet.

THE STUDY
The characteristics of the E1 cooling tower and E2 cooling tower blowdown waters differed greatly in alkalinity, pH, concentration and type of phosphate because different chemical treatment programs are used in each, see Table 2. Therefore, the type of process required to remove phosphate differed for each of the effluent streams to provide adequate flexibility it is desirable that the phosphate removal process be able to treat any combination of effluent streams. The three blowdown streams used in the analysis of the treatment processes were:

• E1 blowdown water
• E2 blowdown water
• A blend of 1/3 E1 blowdown water and 2/3 E2 blowdown water

The study focused on two main processes, a high and a low pH treatment both of which has shown promise during an initial bench scale study.

	Low pH process		High pH process
	160 mg/L of coagulant		160 mg/L of coagulant
	2 mg/L of polymer		2 mg/L of polymer
	1 to 2 % spent lime		200 to 500 mg/L of lime
			1 to 2 % spent lime

Lime dosage varied due to the different buffering capacities of El and E2 blowdown water. The high alkalinity E2 blowdown water required greater dosages of lime than E1 blowdown water in order to maintain the same pH.

In order to meet the AEP maximum allowable concentration of 1 mg/L phosphorous (3 mg/L total phosphate), an orthophosphate concentration of 1.5 mg/L was targeted. This target concentration was determined based on analyses conducted during Associated Engineering's "Phosphate Removal Study". On the basis of 15 samples that were sent to a certified lab, it was determined that the average ratio of orthophosphate to total phosphate was 0.835. Therefore, 3.0 mg/L total phosphate equates to 2.5 mg/L orthophosphate. A target concentration of 1.5 mg/L provides a degree of safety of 1.0 mg/L.

Discussion
Table 3 summarizes the results of all the jar tests, the Associated Engineering pilot plant, and the Densadeg clarifier operation. The successful processes are indicated with a yes (Y) while the unsuccessful processes are indicated with an No (N). For a process to be successful, not only must the target orthophosphate level of 1.5 mg/L be met, but the minimal acceptable rise rate of 12.8 m/hr must be met and sludge production must be less than 5% by volume.

Table 3. Summary of Phosphate Removal Processes

Stream	Study	Process	Alum	Alum, Spent Lime	Alum, Lime	Alum, Lime, Spent Lime
E1 Blowdown	Bench Scale	97 Jar Test PO4	N	X		Y
		98 Jar Test PO4		Y	Y	Y
		97 Pilot PO4
	Densadeg	Clarified PO4		N	Y	Y
		Rise Rate		Y	N	Y
		Sludge		Y	N	Y
		Filtered PO4		Y	Y	Y
E2 Blowdown	Bench Scale	97 Jar Test PO4		Y		Y
		98 Jar Test PO4	Y	Y	Y	Y
		97 Pilot PO4
	Densadeg	Clarified PO4	Y	N		Y
		Rise Rate	Y	N		Y
		Sludge	Y	Y		Y
		Filtered PO4	Y	N		Y
E1 E2 Blowdown	Bench Scale	97 Jar Test PO4	N	N		Y
		98 Jar Test PO4	N	N		Y
		97 Pilot PO4		Y		Y
	Densadeg	Clarified PO4		N		Y
		Rise Rate		N		Y
		Sludge		Y		Y
		Filtered PO4		Y		Y

Discussion of Results. By far the most successful treatment was the high pH process. This

process met orthophosphate target level, the sludge production limitation and the required rise rate for all effluent streams. The low pH process using alum alone, successfully treated E2 blowdown water during the initial parts of the study. However, due to cooling tower process upsets, E2 blowdown water was much more difficult to treat in the latter parts of the study and required the high pH treatment process.

Equipment Performance
The Densadeg clarifier met the target rise rate of 12.8 m/hr which is about three times greater than typical solids contact clarifiers. The greater rise rate for the Densadeg clarifier can be attributed in part to the thickening process in which controlled amounts of sludge are recirculated from the thickener/clarifier vessel into the reactor vessel. This leads to an increase in particle size and weight. Settling velocity will therefore increase based on Stokes' law which states that settling velocity is proportional to particle size and weight. Other factors allowing for higher settling rates are as follows:

• No energy is transferred from the thickener/clarifier turbine to the sedimentation zone. In a normal solids contact unit, energy from the turbine could suspend a sludge blanket.
• The flow does not pass through the concentrated portion of the sludge. This too prevents sludge blanket suspension.
• Tubes settlers and baffled plates assist in the sedimentation process.

One potential disadvantage to the Densadeg clarifier is increased maintenance due to the additional sludge recirculation pump. As well, it was found in the pilot study that the Densadeg clarifier was unable to thicken magnesium hydroxide sludge, but then it is unlikely that any thickener would be able to thicken the same sludge without chemical addition.

Process Challenges
Spikes in hydrocarbon concentration in the blowdown water led to difficulties in meeting the treatment requirements. Hydrocarbon concentrations can build up as the cooling water flows through the heat exchangers. In order to meet treatment requirements, the high pH process was required during periods of increased hydrocarbon concentration.

During the course of this pilot study, blowdown flows occasionally dropped to a small percentage of the proposed design capacity. During cooling tower maintenance, the flow dropped as low as 5% of the proposed design capacity. Even with all the towers in operation, the flows dropped to as low as 10%. Therefore, adequate turn-down capability should be provided.

The most significant process upset that occurred during the pilot study was the precipitation of magnesium hydroxide, Mg(OH)2. Magnesium hydroxide is a low density floc with poor settling properties that results from the removal of magnesium hardness.

Normally when the pH is elevated, addition of lime will remove calcium hardness as calcium carbonate as the primary reaction and magnesium hardness removal will occur to a lesser extent. The calcium carbonate attaches to the lighter magnesium hydroxide floc and assists in settling. However, presence of anti-scaling agents in the blowdown water prevented the precipitation of calcium carbonate.

By inhibiting the precipitation of calcium carbonate, the presence of anti-scaling agents led to greater magnesium hydroxide production. As well, the problem of increased magnesium hydroxide was compounded by the absence of any calcium carbonate to improve the settling properties. It was found during the operation of the Densadeg pilot plant that when more than 200 mg/L of magnesium hardness was removed the following problems were observed:

• Large quantities of sludge were produced, which increased sludge levels within the sedimentation zone up to three metres.
• Increased quantities of sludge were blown down.
• Target rise rates of 12.8 m/hr could not be met. If flow was increased, sludge would pass through the sedimentation zone along with the supernatant.
• Volumetric sludge concentrations within the reactor increased to 23% as compared to the normal 10%.
• Ten minute volume over volume values which determines concentration and thickening properties was recorded at an average of 91%.

With the high magnesium hydroxide precipitation, it was found that colour and phosphate removal increased. However, when considering the above process problems, the excessive precipitation of magnesium hydroxide should be mitigated. During the pilot study, spent lime was added and pH was slightly reduced to mitigate these problems.

Coagulants And Polymers
The pilot study investigated coagulants which would lead to the precipitation of phosphate as a metallic compound. Two types of coagulants were dosed to the various streams, aluminum sulphate (alum) and ferric chloride. It was important to pilot with ferric chloride in the phosphate removal process since this was the optimal coagulant in the softening process. If it was found that ferric chloride worked reasonably well in the phosphate removal process this would eliminate the need for two types of coagulants on site.

The following results were observed from the pilot study:

• Alum showed better results than ferric chloride with regard to turbidity removal.
• Aluminum residuals with the use of alum averaged 0.14 mg/L.
• Iron residuals with the use of ferric chloride averaged 0.36 mg/L.
• Iron residuals decreased as pH increased.

LIME
Lime dosages varied due to the different buffering capacities of the blow down streams. The high alkalinity blow down water required greater dosages of lime than low alkalinity blowdown water in order to achieve the same pH.

Figure 2. Phosphate Removal as a Function of pH

Phosphate removal was dependant on pH. Figure 2 presents the results from the pilot study along with results from a study published by Albertson and Sherwood in 1967.

Spent Lime
Spent lime was obtained from the blowdown line of the existing process water softeners and pumped to the Densadeg pilot plant. The consistency of the spent lime varied due to the dilution from back flush water entering the blowdown system. The average total solids content of the spent lime was determined to be 20%.

Metering of the spent lime at exact concentrations was difficult to achieve. At 20% solids and high hose pump speeds the hose on the suction side of the pump would collapse or the abrasive nature of spent lime would cause hose failure. A progressive cavity pump was also used to meter the spent lime; this failed due to stator wear. Metering by gravity and controlling flow using a valve and rotameter also proved to be inconsistent.

Due to difficulty in metering spent lime, the study team was asked to determine a reasonable solids content of spent lime at which it could be metered. Known concentrations of sludge were diluted to a consistency similar to sludge which could be metered, based on experience. It was determined that the spent lime should be diluted to 10 to 15% solids for metering. As spent lime concentrations from Densadeg clarifiers can reach 30%, a dilution factor of three would be appropriate.

Figure 3. Phosphate Removal with Spent Lime
Because of the difficulties experienced in feeding spent lime, it was eliminated from the pilot plant process for a while. This resulted in process failure. Spent lime serves two purposes in the treatment of phosphate removal; it assists in the adsorption of phosphate, and is used as a weighting agent in the sedimentation process that enables light floc, such as Mg(OH)2, to settle. Figure 3 illustrates the phosphate adsorption capacity of spent lime at various spent lime dosages, without coagulant or polymer addition. Throughout this study it was found that the required spent lime dosage for phosphate removal was approximately 1 to 2% by volume. The precise dosages of spent lime could not be determined.

The availability of spent lime is dependent on the calcium removal in the process water softeners and the flow through these softeners. The available concentration of spent lime which could be metered to the proposed phosphate removal plant will then be a function of the ratio of blowdown water to softened water. This ratio will be affected by the amount of drift from the cooling towers. As drift increases less blowdown is required. Therefore, an increase in drift results in greater concentrations of available spent lime. Table 4 compares the available spent lime concentrations as a function of the above ratio. The calculations were based on the following assumptions:

• raw water calcium 120 mg/L
• softened water calcium 30 mg/L
• calcium carbonate production 90 mg/L
• sludge composition: 95% calcium carbonate
• spent lime total solids: 20%

Softened Water to Blowdown Ratio

Available Spent Lime % Qin

It should be noted that NOVA Chemicals Ltd are actively pursuing drift mitigating measures at the plant site. This will result in blowdown water quantities increasing.

Extraction Sludge
The Densadeg clarifier is designed to produce extraction sludge that is high in solids. Sludge was sampled on seven occasions and sent to a lab for analysis. The highest concentration achieved was 32%. At this concentration the extraction sludge had difficulty flowing through the discharge lines. Sludge was gravity fed into a holding tank and then pumped into a mud tank. At this concentration the sludge pumps operated at a fraction of their capacity.

The sludge from the phosphate removal process was analysed for ICP metals and the important metals are listed in Table 5:

The sludge contains high levels of nutrients, which provides an excellent food source for microbes and algae. Samples of sludge were collected and within one week a variety of biological growths were present. It was also observed that sludge within the tanks went septic within a short period of time; indicative of potential problems with algae growth and odours. For this reason, on-site sludge dewatering should be considered when sludge cannot be immediately disposed of to land application.

Filtration Study
The effluent from the Densadeg was filtered through pilot scale filters at various filtration rates from 5 to 15 m/hr. Turbidity and head loss was monitored continuously while phosphate concentrations were determined by grab samples. The general objective of the study was to determine filter run times and phosphate removal at various filtration rates. Specific objectives of the study were to determine if the high pH process can be replaced with the low pH process plus filtration, and to determine clarified and filtered water characteristics during times of upset.

Ten combinations of blowdown stream, treatment process and chemical treatment were piloted. The results of these combinations are presented in Table 6 which shows clarification alone could not meet the phosphate target level in five of the processes. By filtering the clarified effluent three of these five processes then met the phosphate target levels. As well, three processes that met the average phosphate target level after clarification, had experienced phosphate concentration spikes that exceeded target levels but the spikes were removed through filtration.

Two of the three processes that met phosphate target levels with filtration, but not with clarification, were low pH processes. This indicates that generally a high pH process could be replaced with a low pH process plus filtration. A high pH process would be needed only when blowdown water is more difficult to treat.

The average filtration run times, specific filtration run volumes, and average phosphate removal rates are shown in Table 7. The average run times for filtration rates of 5, 10, and 15 m/hr were 11, 6.75, and 3.8 hours respectively. The largest specific run volume, 67.5 m3/m2, corresponded to a filtration rate of 10 m/hr. In terms of performance, filtration rates of 5 and 10 m/hr averaged 27% phosphate removal, while at 15 m/hr the average phosphate removal dropped to 22%. The choice of filtration rate will depend on economics and the degree of safety required.

Table 6. Summary of Filtration Study

Test #	pH	Chemicals	Clarified OPO4 (mg/L)	Filtered OPO4 (mg/L)	% Reduction thru Filters
E1 Blowdown
2	High	ALP	0.9	0.4	55
3	Low	ASP	2.1	1.4	33
4	High	ASLP	1.0	0.8	20
5	High	FSLP	2.3	1.1	52
6	Low	FSP	2.8	1.2	57
E2 Blowdown
1	Low	AP	0.7	0.3	57
7	Low	ASP	3.1	2.8	10
E1/E2 Blowdown
9	High	ASLP	1.0	0.7	30
10	Low	ASP	3.7	2.3	38
	Failed to meet 1.5 mg/L goal.
A= Alum, F=Ferric Chloride, S=Spent Lime, L=Fresh Lime, P=Polymer

Table 7. Comparison of Filtration Rates

Filtration Rate	5 m/hr	10 m/hr	15 m/hr
Run Time (hours)	11	6.75	3.8
Specific Filtration Run Volume (m3/m2)	55	67.5	57
Average Phosphate Removal (%)	27	27	22

SUMMARY
The pilot study was successful in meeting the objectives set out in the scope of work. Optimal processes were identified along with design criteria for the proposed phosphate removal plant. The conclusions are as follows:

• The high pH process showed the best results regarding phosphate removal, minimal sludge production, and rise rate. The target phosphate concentration of 1.5 mg/L orthophosphate the target volumetric sludge production of 5 %, and the target rise rate of 12.6 m/hr were all met.
• The use of spent lime proved beneficial in the removal of phosphate, increased settling rates and improved sludge thickening. The elimination of spent lime led to process failure. Process failure included large volumes of magnesium hydroxide sludge and low settling rates.
• The metering of spent lime was difficult to achieve.
• The optimal spent lime dosage is 1 to 2%, however the production of spent lime in the softeners is approximately 1%. This percentage will increase as drift increases.
• E1 blowdown water required a more vigorous high pH treatment process compared to E2 blowdown water.
• During the initial part of the study, only alum and polymer were required to reduce phosphate concentrations to below target levels in E2 blowdown water. However, due to cooling tower upsets, the high pH treatment was required during the latter part of the study.
• Alum as a coagulant resulted in lower turbidity as compared to ferric chloride. Also, metal residual concentrations were lower with alum.
• The Densadeg clarifier thickened the extraction sludge up to 32%, which will require dilution to achieve pumpable sludge.
• The extraction sludge contained high levels of nutrients, which provide an excellent food source for microbes and algae. It was observed that the extraction sludge in the holding tanks and sample containers went septic within a short period of time.
• Softening of the blowdown water was inhibited by the presence of anti- scaling agents.
• The use of filters further reduced phosphate levels by an average of 27%. Filtration also reduced phosphate to target levels in three of the five processes where clarified effluent did not meet target levels.
• At filtration rates of 5 and 10 m/hr, phosphate removal was not significantly different; corresponding filter runs were 11 to 6.75 hours. When the rate was further increased to 15 m/hr, filter run times dropped to 3.8 hours and average phosphate reduction dropped to 22%. The maximum specific filter run volume was 67.5 m3/m2, at 10 m/hr filtration rate.

ACKNOWLEDGEMENTS
The following personnel are acknowledged for their respective contributions to this study:

1. Nandan Vani, P.E., Bantrel Inc., Environmental Engineer, who developed the study programme.

2. C.J. Lee, P.Eng., NOVA Chemicals Ltd., Process/Environmental Engineer, who directed and managed the programme.

The author is indebted to these people for their guidance and contribution to this paper.