Mauritania - Guelbs and SNIM Projects

Document of The World Bank FOR OFFICIAL USE ONLY Report No. 10781 ISLAMIC REPUBLIC OF MAURITANIA PROJECT PERFORMANCE AUDIT REPORT GUELBS IRON ORE PROJECT (LOAN 1747-MAU) PROGRAM PERFORMANCE AUDIT REPORT SNIM REHABILITATION PROJECT (LOAN 2643-MAU) JUNE 26, 1992 MICROFICHE COPY Report No. 1078q-MAU Type: (PER) NAMISATO, / X31678 / T9105/ OEDD2 Operations Evaluation Department This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. CURRENCY EQUIVALENT 1979 1985 May 1991 Currency Unit Ouguiya (UM) US$1.00 UM 45.85 UM 77.07 UM 83.20 UM 1.00 US$0.022 US$0.013 US$0.012 WEIGHTS AND MEASURES EQUIVALENTS 1 kilogram (kg) - 2.205 pounds (lb) 1 metric ton (mt) - 2,205 pounds (lb) 1 meter (m) - 3.281 feet (ft) 1 cubic meter (M) = 33.315 cubic feet (ft3) 1 kilometer (km) - 0.62 mile (mi) 1 square kilometer (kM2) - 0.386 square mile (sq mi) ABBREVIATIONS AND ACRONYMS ADB - African Development Bank AFESD = Arab Fund for Economic and Social Development ARMICO = Arab Mining Company CCCE - Caisse Centrale de Coop6ration Economique COMINOR = Complexe Minier du Nord EEC - European Economic Community EIB - European Investment Bank FCB - Five Cail Babcock FED - Fonds Europ6en de D6veloppement IRSID - Institut de Recherche de la Sid6rurgie KFAED = Kuwait Fund for Arab Economic Development KFTCIC - Kuwait Foreign Trading, Constructing and Investment Co. MIFERMA - Mines de Fer de Mauritanie MSI = Mineral Services Incorporated OECF - Overseas Economic Cooperation Fund (Japan) OED = Operations Evaluation Department PCR - Project Completion Report PPAR - Project Performance Audit Report SAR - Staff Appraisal Report SNIM - Soci6t6 Nationale Industrielle et Mini6re SOCOMINE - Socidtd de Coop6ration Mini6re et Industrielle SOFREMINES - Socidt6 Francaise d'Etudes Mini6res SOFRESID = Soci6td Francaise d'Etudes de la Sid6rurgie SGTE = Soci6td G6nerale de Techniques et d'Etudes TPY - Tons per year FISCAL YEAR January 1 - December 31 FOR OFFICIAL USE ONLY THE WORLD BANK Washington, D.C. 20433 U.S.A. (Vice of Directnh-General , Operations Evaluation June 26, 1992 MEMORANDUM TO THE EXECUTIVE DIRECTORS AND THE PRESIDENT SUBJECT: Project Performance Audit Report on Mauritania - Guelbs Iron Ore Project and SNIM Rehabilitation Project (Loans 1747-MAU and 2643-MAU) Attached, for information, is a copy of a report entitled "Project Performance Audit Report on Mauritania - Guelbs Iron Ore Project and SNIM Rehabilitation Project (Loans 1747-MAU and 2643 -MAU)," prepared by the Operations Evaluation Department. Attachment This document has a restricted distribution and may be used by recipients only In the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. FOR OFFICIAL USE ONLY PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA GUELBS IRON ORE PROJECT AND SNIM REHABILITATION PROJECT (Loan 1747-MAU and Loan 2643-MAU) TABLE OF CONTENTS Page No. Preface.............................................. j Basic Data Sheets...................................................... iii Evaluation Summary............ ....................................... ix I. BACKGROUND................................... 1 II. PROJEOT OBJECTIVES, DESCRIPTION AND DESIGN...................... 2 A. Objectives....................................... ....... 2 (1) The Guelbs Iron Ore Project (1979)..................... 2 (2) The SNIM Rehabilitation Project (1985)................... 3 B. Description and Design......... ........................... 3 (1) The Guelbs Iron Ore Project.............. ....... 3 a) Technical aspects.............. 4 b) Project Cost and Financing..... .................... 8 c) Sales and Profitability............-oo ...... ....... 9 (2) The SNIM Rehabilitation Project.......... ...... 11 III. IMPLEMENTATION... . ..................................e-e- 15 A. Guelbs Iron Ore Project.........o ............... ..... 15 (1) Implementation Delays.... ........................ 16 a) Delays in Effectiveness........o..o.oo ..... o. 16 b) Delays Oue to Procurement Problems................... 16 c) Delays in Construction............. ......... .... 17 (2) Start-up Technical Problems......... ................... 18 a) Dust Problem.............................. .... 18 b) Equipment Deficiencies................................ 18 c) Grinding Mills............. . .... ... ..........19 d) Separation Process................................ 20 B. SNIM Rehabilitation Project.... ............ 20 IV. RESULTS......... 22 A. The Guelbs Iron Ore Project.. ............... 22 B. The SNIM Rehabilitation Project..... .................. .27 V. SUSTAINABILITY... ............................ ..... 29 VI. FINDINGS AND LESSONS OF EXPERIENCE...... ............... ..... o 32 This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. Table of Contents (cont'd) Page No. ATTACHMENTS I. Response from the Ministry of Planning ......................... 37 II. European Investment Bank's Comments ......................000.. 38 AN : Mauritania - SNIM Historical Income Statement ................ 45 PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA GUELBS IRON ORE PROJECT AND SNIM REHABILITATION PROJECT (Loan 1747-MAU and Loan 2643-MAW) I. PREFACE 1. This is a Project Performance Audit Report (PPAR) on Loan 1747-MAU in the amount of US$60 million, and Loan 2643-MAU in the amount of US$20 million to Socidt6 Nationale Industrielle et Mini&re (SNIM). Both loans were guaranteed by the Government of Mauritania. Loan 1747 was approved on July 12, 1979 and became effective on December 31, 1980. Loan 2643 was approved on December 17, 1985 and became effective on March 19, 1986. The original closing date of June 30, 1983, for Loan 1747 was postponed to December 31, 1984, while the Loan 2643 was closed on December 31, 1988 as originally planned. Both loans were fully disbursed. 2. The PPAR was prepared by the Operations Evaluation Department (OED) and is based on the Prol ect Completion Report (PCR) prepared by the Africa Regional Office of the Bank. The PPAR is based on the PCR, the Staff Appraisal Reports (SAR), the President's Reports, the loan documents, the summary of the Executive Directors' meetings at which the Projects were considered, a study of the projects' files and discussions with Bank staff. An OED mission visited Mauritania in April 1991. That mission reviewed the effectiveness of the Bank's assistance with SNIM's managers and Government's officials. Their kind cooperation and valuable assistance in the preparation of this report is gratefully acknowledged. 3. The PCR provides a good account and assessment of the projects experience, and discusses the performance of the Bank and of the borrower. The PPAR elaborates on particular aspects such as the critical issues of concern that led to the genesis of these loans in conjunction with the need to maintain a viable mining sector essential to Mauritania's economy; evaluates the problems encountered in implementing the various components of the projects; and ascertains the key factors that determined the loans' outcome and sustainability of the projects. The PPAR, then, draws further lessons from the projects' experience. 4. The draft PPAR was sent to the Borrower and the cofinancing agencies for comments. The comments received from the Government (Attachment I) and the European Investment Bank (Attachment II) are included in the PPAR. L Project Completion Report, Mauritania-Guelbs Iron Ore Project (Loan 1747- MAU) and SNIM Rehabilitation Project (Loan 2643-MAU), Report No. 9487, dated April 19, 1991. PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA GUELBS IRON ORE PROJECT (Loan 1747-MAU) BASIC DATA SHEET LOAN POSITION (amounts in US$ million) As of March 31, 1992 Loan Orisinal Disbursed Cancelled Repaid O.totandine 1747 60.00 60.00 - 45.00 15.00 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS FYS0 F-Y83 FY85 Appraisal Estimate (US$ million) 3.00 60.00 Actual (US$ million) - 32.80 60.00 Actual as I of Appraisal (2) - 54.7 100.00 Date of Final Disbursement May 30, 1985 PROJECT DATES Oriainal Actual Identification 01/15/77 Appraisal 10/03/77 Negotiations 02/28/79 Board Approval 07/12/79 Signing 12/14/79 Effectiveness 12/31/80 Project Completion 01/01/83 03/31/85 Loan Closing 06/30/83 12/31/84 STAFF INPUTS (Staff veeks) Preparation Agraleal Nepotiations Supervision Comgletionu Tota FY77 49.6 49*6 FY78 25.7 55.5 81.2 FY79 37.0 36.3 73.4 FY80 1.0 30.4 31.4 FY81 13.4 13.4 FY82 38.8 38.8 FY83 20.5 20.5 FY84 20.8 20.8 FY85 6.9 6.9 FY86 0.3 0.3 FY87 2.7 2.7 MLAL 75.3 92.5 37.3 133.7 338.9 Staff vaeks included in staft inputs for SNIN Rehabilitation Project. MISSION DATA MonthiYear No. Weeks No. Persons Staff-Weeks Preparation 01/77 3 7 21 Appraisal 10/77 3 3 9 Post Appraisal 08/78 1 3 3 Supervision 1 10/79 1 3 3 Supervieon II 04/80 1 4 4 Supervision III 12/80 1 2 2 Supervision IV 09/81 1 3 3 Supervision V 04/82 2 3 6 Supervision VI 10/82 2 3 6 Supervision VII 04/83 1 3 3 Supervision VIII 12/83 1 3 3 . V . OTHER PROJECT DATA Related Prolects Borroweras MIFERMA (Loan 0249); SHIM (Loar 1747) and Government of Mauritania (Credit 2166) Project: MIFERMA - Iron Ore Loan No.t 0249-MAU Amount: US$66.0 million Approval Date: 03/15/60 Project: SHIN Rehabilitation Project Loan No.t 2643-HAU Amounts US$20.0 million Approval Dates 09/15/85 Project: Public Enterprise Sector Adjustment Programs Credit No: 2166-MAU Amount: US$40.0 million Approval Date: 06/26/90 PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA SNIM REHABILITATION PROJECT (Loan 2643-MA) BASIC DATA SHEET LOAN POSITION (amounts in US$ million) As of March 31, 1992 Loan Original Disbursed Cancelled Revaid Outstaa4ing 2643 20.00 20.00 - 5.01 14.99 CUMULATIVE ESTIMATED AND ACTUAL DISBURSEMENTS FY87 FY88 FY89 Appraisal Estimate (US$ million) 6.00 15.00 20.00 Actual (US$ million) 10.00 20.00 20.00 Actual as 2 of Appraisal (2) 167.00 133.30 100.00 Date of Final Disbursement September 15, 1987 PROJECT DATES Orisinal Actual Identification 03/01/84 Appraisal 04/25/85 Negotiations 10/15/85 Board Approval 12/17/85 Signing 01/27/86 Effectiveness 03/19/86 Project Completion 06/30/88 06/30/88 Loan Closing 12/31/88 12/31/88 - vii - STAFF INPUTS (Staff weeks) FY84 V85 FY86 FY87 FY88 FY89 yg0 FY91 TIOAL Preparation 45.7 23 9 69.6 Appraisal 16.0 16.0 32.0 Negotiations 8.7 8.7 Supervision 5.4 18.6 15.3 1.3 0.3 40.9 Completion 5.8 0.2 6.0 TOTAL 45.7 39.9 30.1 18.6 15.3 1.3 6.1 0.2 157.2 MISSION DATA l4ntlesrg No. Weeks No. Persons Staff-Weeks Preparatio 10/84 1 3 3 Appraisal 08/85 2 2 4 Post Appraisal 09/85 1 2 2 Supervision 1 05/86 1 1 1 Supervision II 10/86 1 1 1 Supervision III 12/86 1 1 1 Supervision IV 05/87 1 2 2 Supervision V 06/88 2 2 4 viti - OTHER PROJECT DATA Related Proiects Borroverst MIFERMA (Loan 029); SHIM (Loan 2643) and Government of Mauritania (Credit 2166) Projects HIFERMA Iron Ore Loan No.x 0249-MAU Amount: US$66.0 million Approval Dates 03/15/60 Projects Guelbe Iron Ore Loan No.: 1747-MAU Amounts US$60.0 million Approval Date: 07/12/79 Project: Public Enterprise Sector Adjustment Program Credit No.t 2166-MAU Amount: US$40.00 million Approval Date: 06/26/90 -ix- PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA GUELBS IRON ORE PRJECT AND SNIM REHABILITATION PROJECT (Loan 174/-MAU and Loan 2643-MAU) EVALUATION SUMMARY 1. Bank involvement in Mauritanian growth. In 1974, mining directly iron ore dates back to 1957, when accounted for 29% of Mauritania's Socidtd Anonyme des Mines de Fer de G.D.P. at factor cost. This was the M&uritanie (MIFERMA), a company owned year when iron ore production reached by European interests, requested the a peak of 11.6 million tons. With the Bank to finance the exploitation of decrease of output to an average level the Kedia high-grade iron ore of 9 million tons in 1975-1987, and deposits, 650 kilometers northeast of the growth of other subsectors such as Nouadhibou, the main port. The Bank fish processing and services, the loan, made in 1960, anounted to US$66 share of mining in G.D.P. declined to million, out of a total project cost 11-12% in most of the 1980s. In 1990, of US$190 million. The exploitation mining accounted for about 15% of started in 1963. (para. 1) G.D.P. as a result of increased ore output (11 million tons) and relative 2. In 1979, the Board approved a decreases in the primary sector and in US$60 million loan to SNIM - - the fish processing. (para. 5) state corporation, which had taken over the MIFERMA operation in 1974 -- 5. The financing of the Guelbs to finance part of the first phase of Project in 1979 by the Bank and other the Guelbs Iron Ore Project ("guelbs" donors was seen as a sign of support consist of a number of iron ore to the new government installed in outcroppings). (para. 2) 1978 after a coup d'etat and in the midst of guerilla attacks on the 3. In 1925, another US$20 million railway, which made sometimes loan to SNIM was approved with the difficult the evacuation of iron ore objectives to continue to improve to the port of Nouadhibou. The SNIM's management, reduce its importance of maintaining iron ore production costs and thus secure the production was judged so vital for the financial and economic viability of Mauritanian community that the the company. (para. 3) decision was made to go ahead with the project despite the political and 4. The long and deep Bank security risks involved. (para. 9) involvement in Mauritanian iron ore is due to the fact that it is a mainstay 6. Similarly, when faced with an of Mauritania's economy. Iron ore untenable economic and financial mining development throughout the situation, the Mauritanian government 1960's was the principal determinant adopted, in September 1985, a of the country's rapid economic comprehensive Economic and Financial -x - Recovery Program (PREF) for the 1985- of 14 million tons per year, and to 88 period, such Recovery Program was assure the continued financial reinforced by the 1985 SNIM viability of SNIM. (para. 11) Rehabilitation Project. Clearly, given the importance of iron ore 8. The main technical feature of mining in the country's economy, a the project was a large bonification link had to be made between macro- plant using a dry, magnetic separation economic programs and the need to process. This plant was necessary maintain a viable mining industry. In since the low iron content of the 1990, incomplete public enterprise Guelbs ore (38% Fe) had to be reforms were again addressed in the concentrated to be marketable. context of a Public Enterprise SECAL Although the technology used was which benefitted SNIM to a major tested, some independent experts extent and carried the reforms expressed doubts about the feasibility considerably further than was of the dry separation process. originally envisaged under the PREF (paras. 22-31) and the 1985 SNIM Rehabilitation Project. (para. 10) 9. Despite projected rather low financial and economic profit.-bility Project Objectives and Design (5.4% and 11.8%, respectively), the overwhelming concern of the main A. Guelbs Iron Ore Project shareholder, the Mauritanian Government, was the suppozcdly high 7. The main objective of the economic benefit of the project, Guelbs Iron Ore Project was to open extending the life of the :nine, and new mines and constructing thus developing in the long term the bonification facilities, in order to major productive sector of the replace the declining production of country's economy and the only the existing Kedia high grade (64% Fe) sustainable activity in the northern ore bodies, which were then expected part of Mauritania. This concern was to be exhausted by 1991. The project also widely shared by some Bank staff was to be implemented in two phases (who insisted on the lack of with the objective of increasing alternative large viable projects in SNIM's production capacity from 11.5 Mauritania) and by other lenders, in million tons to about 14 million tons particular the Arab countries. (para. per year by 1984. The first phase, 54) which was to only partially replace the Kedia mining operations, was to B. The SNIM Rehabilitation Project bring a new low grade (38% Fe) ore (1985) body, El Rhein (located 22 km northeast of Zouerate), into 10. The objectives of this project production in 1983. The second phase were to consolidate and further was to open other low grade ore mines advance SNIM's program for managerial at Oum Arwagen (12 km east of El and technical improvements during the Rhein). These were to come into years 1986-88 and in particular were production in 1988, to replace the to (i) help SNIM continue to improve final ores from Kedia. The Bank loan its management and reduce its was to finance the first phase but it production costs; and (ii) assist with was considered imperative that the the necessary finsucial restructuring second phase be forthcoming, both in of SNIM by transforming part of its order to maintain the projected output short-term debt/overdrafts into long- - xi - term debt. These measures were with the core beneficiation plant to necessary due to the fact that SNIM's be completed by March 1982. However, financial viability had been the latter was completed in May 1984, threatened by the combined effort of i.e., more than two years later than adverse iron ore price and market anticipated. Erection of the power conditions, by lower than expected plant was also delayed by 20 months, results of the Guelbs project and high while mining and railroad components production costs. (para. 12) were implemented ahead of schedule (six and two months respectively). 11. The project included (para. 80) replacement of equipment, major overhauls of equipment, provision of 15. In 1985, production was only tools and instrumentation for about 0.9 million tons as compared to workshops and technical assistance for the 6.2 million tons originally a new management information system, estimated. Production rose to 1.4 mine planning, railway maintenance and million tons in 1986-1987 and to 2 spare parts management. (para. 67) million in 1988-1990, but never reached the expected level of 12.5 12. The projected financial and million tons in 1990 (including Phase economic rates of return were 21% and II of the project, which was never 25% respectively, much higher than for undertaken). (para. 81) the Guelbs project and were calculated on an incremental production basis, 16. The delays, which occurred i.e. it was assumed that the during project implementation, are Rehabilitation Project could be attributable partly to suppliers and credited with a level of production of partly to the Engineer's lack of one million tons per year higher than experience in executing projects of in the without-project case. (paras. this complexity. There were three 75-77) main types of delays: (a) delays in loan effectiveness, (b) delays in 13. The Bank ar1 some other procurement, and (c) delays in European lenders wanted to pursue for construction. (para. 82) the Guelbs project the same "enclave" project approach which had been used 17. Serious technical problems for MIFERMA. This approach was which delayed the plant start-up and assisted by loans in relatively hard have still not been fully overcome terms when the project was known to be were the dust problem, the equipment only marginally financially viable. deficiencies, the grinding mill This led to high debt service and capacity problem and the separation finally to massive debt relief. The process. (para. 94) 1990 IDA credit was granted with the knowledge it would be largely used to B. SNIM Rehabilitation Project help SNIM repay its earlier loans. (para. 155, ) 18. The Rehabilitation Project was implemented much more satisfactorily, Implementation although a number of problems faced during the Guelbs Project A. Guelbs Iron Ore Project implementation phase have not yet been fully solved. (para. 102) 14. The Project was to be implemented between end-1979 and 1983 19. Modifications and replacements - xii have been ongoing to the present, objectives which were to: (i) help particularly in view of the fact that SNIN to improve its management and the 1985 loans for SNIM Rehabilitation reduce its production costs; (ii) were followed in 1987 by other loans, assist with the necessary financial totalling US$50.8 million, from Arab restructuring of SNIM by transforming and French lenders as well as by the part of its short-term debt/overdrafts EC, EIB, and ADB (the Bank did not into long-term debt, have been participate). A US$40 million IDA substantially achieved. (para. 122) credit approved in 1990 is being partly used (together with other 23. The recalculated financial rate resources provided by Arab, French and of return of the SNIM Rehabilitation Japanese sources) to further help SNIM Project has been calculated at 21%, or in completing the Guelbs technical and the same as at appraisal in 1985. The managerial action program. (para. recalculated economic rate is 19%, 104) only slightly lower than estimated at appraisal (25%). Substantial cost Results savings have been achieved as well as lower operational costs. (para. 131) A. Guelbs Iron Ore Project 24. Rates of return of the joint 20. Iron ore production has been projects are still highly much lower than anticipated at unsatisfactor, i.e. -21 (financial) appraisal. Design problems and severe and 01 (economic return). In order to implementation difficulties have achieve a satisfactory rate of return, prevented the Project to reach its iron ore prices would have to increase production objectives. The problem by almost 100% or the investment would lies with the Guelbs bonification have to be devalued to 10% of its plant. Dust and equipment actual cost. (para. 133) deficiencies have been major factors in precluding the plant to produce gustainability more than two million tons in 1990, i.e. only 33% of the nominal capacity. 25. Present forecasts are that the (para. 106-108) Guelbs iron ore mine will slowly increase production from 2 million 21. The recalculated financial tons in 1990 to 4 million tons in 1996 return of the Guelbs project was and remain at that level up to year negative (-10%) compared to the 2000. Although production would be already low 1979 appraisal estimate of more than one-third below the original 5.4%. Such a negative result was due forecast (Phase I), it would represent to the severe technical problems of a doubling in output between 1990 and the plant, resulting high production 1996 because of substantial costs and the much-lower-than- investments already made and ongoing estimated iron prices. The major overhaul of equipment. (para. recalculated economic rate of return 134-135) was a negative 5% compared with 11.8% at appraisal. (paras. 115-117) 26. Such increase in production would lead to a decrease in operating B. The SNIM Rehabilitation Project costs (about 15%), but this would not be sufficient to offset the combined 22. Results of this Project have effect of: (i) slow but continuous been gererally positive. The main decline in iron ore prices now - xii - forecast by most experts in the 1990s Mauritania, because of its capacity to due to poor prospects in the steel generate foreign exchange for the industry, and (ii) high financial Mauritanian economy, its ability to charges resulting from a heavy debt transfer technology to the whole burden and depreciation of costly industry sector, and to create investments. As a result, the Guelbs employment in the modern sector. project would generate enough revenues (para. 140) to cover operating costs but would continue to have a negative net income LSssons of ExRerience in the foreseeable future. (para. 136) 30. The following lessons can be 27. Fortunately, prospects for derived from the Cuelbs and SNIM SNIM, as a whole, are somewhat Rehabilitation Projects (para. 155): brighter. Because the Kedia mines have been able to increase production (a) Large mining expansion projects in 1989-90 and sell ore at somewhat have to be undertaken only after a better prices, SNIM's financial careful iron exRloration program has performance has improved and net been designed and executed. No profits were made in the last two preconceived assumptions such as the years. Prospects are that SNIM should fact that previous exploration work remain profitable in 1991-92 but again has been undertaken in earlier years incur losses in 196-2000 partly due and no rich ore was found, should be to the relatively high depreciation taken as an excuse for not conducting charges for both the Cuelbs and the further exploration work. Mining new high grade ore M'haoudat project, projects are usually highly risky and but also because the average cost of continuous, careful exploration goods sold is expected to increase due program is a way to minimize such to the higher share of Guelbs risks. The Guelbs Project is another production in SNIM's total output. example of the danger for geologists (paras. 137-138) to adopt excessively simplified conclusions. 28. Although SNIM is expected to have a negative net income during the (b) The same applies to such an second half of the 1990s, the longer important resource as water, term financial outlook should improve particularly in a desertic since net losses would progressively environment. Systematic search for decline due to reduced interest additional water resources, which were payments and depreciation charges found later, would have totally (except M'haoudat) and represent not changed technology choices for the more than 3.5% of SNIM's total Guelbs Project. revenues in year 2000. Moreover, financial projections have been (c) Reliance on pilot testing and prepared in constant dollar terms expert technical committees is no which seems to somehow overstate the safeguard against the risks connecte weight of SNIM's debt service with RrototyRe Rlants. The decision obligations. (para. 139) to go ahead with such plants should not be made without full consideration 29. The likely sustainability of of all elements involved: ore the SNTE Company, if not of the Guelbs hardness; scaling up factors from Project as such, is important since pilot plant testing; availability of SNIM remains a valuable asset for suitable, reliable equipment; and - xiv - experience with other prototypes profitability obtained by using lower elsewhere. In principle, the use of price projections. neu untested technology should be avoided, particularly in a difficult, (g) Consideration should be given desertic environment such as the to the high risks resulting fro Sahara. Even today, experts remain foreign exchange fluctuations. Such divided on the question as to whether risks may prove very high indeed the magnetic separation dry process depending on the denomination of the was the best available. debt in various currencies and on the origin of equipment purchased. (d) When selecting equipment, Allowance should be made for such maximum weight should be given to fluctuations in sensitivity analysiz sturdiness even if capital costs seem and project risks calculations. relatively high, whenever this is compatible with the project's (h) A coherent link has to be finances. The Guelbs Project was established betwe3n terms of lending completed at a cost lower than on one hand and the financial originally estimated partly because prospects of the Project on the other too much stress may have been put on hand. Lenders' understandable desire locating cheaper suppliers but the to be repaid fully and in time can real cost resulting from equipment only be frustrated if the burden of deficiencies has been extremely high. debt becomes unbearable for the borrower. Financing must be adapted (e) Experienced operators from a to the characteristics of the project. similar industrial plant should be sought to design and implement such a (i) Financing mining projects with complex project. Before relying on a doubtful financial and economic engineers with integrity and justification should be avoided. An professional competence in iron ore overly pessimistic view of the mining but not in complex bonification country's economic prospects may lead technical process, alternative to major mistakes. The decision to go solutions such as hiring a specialized ahead with the Guelbs Project was made engineering firm should be carefully just before the fisheries sector evaluated. It is to be regretted that started to develop to the point of only after several years of bad becoming more important in a few years experiences, SNIM hired a than the mining sector itself. Fish manager/engineer, who had a key role exports have provided substantial in a similar Canadian project. foreign exchange receipts to Mauritania. The decision to finance (f) When a project (such as Guelbs) the Guelbs Project did not prove fully is known at appraisal as having a very justified from a balance of payments marginal rate of return, extra- point of view. A more global and conservative price assumptions should effective approach would have been to be made. Sensitivity analysis should evaluate this important project in a show results obtained by using price wider macro-economic context, taking forecasts below the official Bank's into account the long-term prospects escimates and the final decision to of the economy as well as potential finance the project should be made in development of all relevant sectors. giving utmost consideration to PROJECT PERFORMANCE AUDIT REPORT ISLAMIC REPUBLIC OF MAURITANIA GUELBS IRON ORE PROJECT AND SNIM REHABILITATION PROJECT (Loan 1747-MAU and Loan 2643-MAU) I. BACKGROUND 1. Bank involvement in Mauritanian iron ore dates back to 1957, when Soci6td Anonyme des Mines de Fer de Mauritanie (MIFERMA), a company owned by European interests, requested the Bank to finance the exploitation of the Kedia high-grade iron ore deposits, 650 kilometers northeast of Nouadhtbou, the main port. The Bank loan, made in 1960, amounted to US$66.0 million, out of a total project cost of US$190 million. The exploitation started in 1963. 2. In 1979, the Board approved a US$60 million loan to SNIM -- the state corporation, which had taken over the MIFERMA operation in 1974 -- to finance part of the first phase of the Guelbs Iron Ore Project ("guelbs" consist of a number of iron ore outcroppings). 3. In 1985, another US$20 million loan to SNIM was approved with the objectives to continue to improve SNIM's management, reduce its production costs and thus secure the financial and economic viability of the company. 4. In 1990 a Development Credit of US$40 million equivalent was approved to support the extension of the Government's structural adjustment program aimed at deepening the reforms in the public enterprise sector. SNIM was the largest beneficiary. The program included major financial restructuring measures, not only to ensure that the financial burden resulting from the difficulties of the Guelbs iron ore project was substantially reduced by providing further debt relief, but also to enable SNIM to maintain its current production levels in the long-run. 5. The long and deep Bank involvement in Mauritanian iron ore is due to the fact that it is a mainstay of Mauritania's economy. Iron ore mining development throughout the 1960's was the principal determinant of the country's rapid economic growth. In 1974, mining directly accounted for 29% of Mauritania's G.D.P. at factor cost. This was the year when iron ore production reached a peak of 11.6 million tons. With the decrease of output to an average level of 9 million tons in 1975-1987, and the growth of other subsectors such as fish processing and services, the share of mining in G.D.P. declined to 11-12% in most of the 1980s. In 1990, mining accounted for about 15% of G.D.P. as a result of increased ore output (11 million tons) and relative decreases in the primary sector and in fish processing. 6. Iron ore mining has a substantial indirect impact on G.D.P. Many industrial and construction enterprises work r nly, or exclusively, for SNIM. Mining creates directly or indirectly local addet ralue, which has been estimated to about 10% of the value added in the industrial, construction, and services sectors, in the mid-1970s. This, in turn, represented about 10% of G.D.P. -- -2- implying a contribution of some 31% in toto (the direct contribution averaged 21% in 1973-76). In addition, although SNIM's poor financial situation has brought no or little budgetary benefits for the country, the company indirectly contributes several million dollars to the Government's budget in the form of workers' income tax. 7. In the 1970s, iron ore exports accounted for 80% of Mauritania's total value of exports. With the major subsequent expansion in fish exports and the decline in ore prices, the share of ore exports declined to a low of 22% in 1987. However, due to higher volume and prices of ore exports, and lower fish exports, the share of iron ore in total merchandise exports rose to 41% in 1989 and 49% in 1990. 8. Iron ore operations currently provide about 4,500 direct jobs. These, in turn, sustain the entire population of Zouerate (50,000 people) where the mines are located, along with a major proportion of that of the port of Nouadhibou (70,000), from which iron ore is exported. Furthermore, mining indirectly accounts for a substantial proportion of the construction and services sector labor force. 9. The financing of the Guelbs Project in 1979 by the Bank and other donors was seen as a sign of support to the new government installed in 1978 after a coup d'etat and in the midst of guerilla attacks on the railway, which made sometimes difficult the evacuation of iron ore to the port of Nouadhibou. The importance of maintaining iron ore production was judged so vital for the Mauritanian community that the decision was made to go ahead with the project despite the political and security risks involved. 10. Similarly, when faced with an untenable economic and financial situation, the Mauritanian Government adopted, in September 1985, a comprehensive Economic and Financial Recovery Program (PREF) for the 1985-88 period, such Recovery Program was reinforced by the 1985 SNIM Rehabilitation Project. Clearly, given the importance of iron ore mining in the country's economy, a link had to be made between macro-economic programs and the need to maintain a viable mining industry. In 1990, incomplete public enterprise reforms were again addressed in the context of a Public Euterprise SECAL which benefitted SNIM to a major extent and carried the reforms considerably further than was originally envisaged under the PREF and the 1985 SNIM Rehabilitation Project. II. PROJECT OBJECTIVES, DESCRIPTION AND DESIGN A. OBJECTIVES (1) The Guelbs Iron Ore Project (1979) 11. The main objective of the Guelbs Iron Ore Project was to open new mines and construct bonification facilities, in order to replace the declining production of the existing Kedia high grade (64% Fe) ore bodies, which were then expected to be exhausted by 1991. The project was to be implemented in two phases with the objective of increasing production capacity from 11.5 million tons to about 14 million tons per year by 1984. The first phase, which was to -3- only partially replace the Kedia mining operations, was to bring a new low grade (38% Fe) ore body, El Rhein (located 22 km northeast of Zouerate), into production in 1983. The second phase was to open mines at Oum Arwagen (12 km east of El Rhein). These were to come into production in 1988, to replace the final ores from Kedia. The Bank loan was to finance the first phase but it was considered imperative that tLe second phase be forthcoming, both in order to maintain the projected output of 14 million tons per year (tpy), and to assure the continued financial viability of SNIM. (2) The SNIM Rehabilitation Project (1985) 12. The objectives of this project were to consolidate and further advance SNIM's program for managerial and technical improvements during the years 1986-88 and in particular were to (i) help SNIM continue to improve its management and reduce its production costs; and (ii) assist with the necessary financial restructuring of SNIM by transforming part of its short-term debt/overdrafts into long-term debt. These measures were necessary due to the fact that SNIM's financial viability had been threatened by the combined effect of adverse iron ore price and market conditions, by lower than expected results of the Guelbs project and high production costs. B. DESCRIPTION AND DESIGN (1) The Guelbs Iron Ore Project 13. Even during MIFERMA's initial approval of the Kedia project in the late 1950's, it was apparent that although the mountains contained a satisfactory reserve, it was limited. The sponsors, cherefore, intended, from the outset, to explore and develop the so-called "guelbs" (a number of iron ore outcroppings nearby), in order to continue to make use of the large infrastructure investments -- notably the 650 km railway and port facilities -- once the Kedia ore was exhausted. 14. Since start-up in 1963, production had reached a peak production level of 11.4 million tons in 1974, well above the 6 million tpy output originally envisioned. The nationalization of MIFERMA in 1974 did not seriously disrupt iron ore deliveries, and following successful negotiations with the former European shareholders -- which led to a generous US$90 million nationalization compensation - - a normal business relationship was resumed with Socidtd Nationale Industrielle et Minibre (SNIM), the state corporation which had taken over the MIFERMA operations. 15. After 1974, sales volume began to decrease (as low as 6.5 million tons in 1978), as the first effects of the recession of the steel industry in the major producing countr!-is were felt. Average F.O.B. prices for Kedia ore fell from US$15.6/ton in 1975 to US$11.8 in 1979, and SNIM started to make losses (in constant 1977 terms). 16. It was, however, anticipated that Kedia ore prices would increase again from 1980 on and reach US$15.2/ton in 1982 (in constant 1977 terms) to remain at that level up to 1991. This optimistic price forecast was based on the assumption that a fairly balanced world supply/demand situation was to be -4- expected through the 1980's. It was also believed (SAR Guelbs Iron Ore Project, June 20, 1979) that the Guelbs Project would not experience substantial marketing problems, since the increment in iron ore supply which the project was to introduce on to the world market was rather small (3-4 million tpy, or less than 1% of total world net ore exports). 17. Another important consideration was the low iron content of the Guelbs ore (the mean mining grade was established as being about 38% Fe). This .eant that ore had to be concentrated to be marketable. The concentrates to be produced by the Guelbs were to be of the "sinter" feed type or "fine" ore with a predictable granularity. Such sinter feed concentrates are mixed with coke in blast furnaces to produce pig iron. They are in competition with other products such as pellets, lumpy ores, and natural fines. However, technical studies showed that the advantage of sinter lied in the fact that is can be self-fluxing, thus reducing coke consumpLion and leading to productivity gains in the blast furnace and substantial cost savings per ton of pig iron produced. A premium demand was therefore assumed for Guelbs sinter feed. a) Technical aspects Ore Reserves and Mineralo&y 18. Proven ore reserves at El Rhein and Oum Arwagen were estimated at 285 and 101 million tons respectively. To this must be added the remaining 96 million tons reserves of ths Kedia mines, to yield a total proven reserve inventory of 481 million tons. Probable extensions of these ore bodies could have added another 180 million tons. 19. The overall mineralogy of the Guelbs ore is simple. It consists of a facies of coarse and fine magnetite, as well as coarse and fine hematite (referred to by SNIM as oxidized ore), with a ratio between the two of 2:1 respectively. The mineralogical studies undertaken to establish the first parameters for the pilot plant flow sheet were extensive. During the period 1971-75, both laboratory-type and semi-industrial tests were carried out in Europe, Canada, the U.S.A., and Japan. 20. To supplement the earlier laboratory work, a pilot plant was erected in Zouerate in 1974 to: (i) separately treat all major types: (ii) determine the permissible mixtures between magnetic and oxidized ores, so as to arrive at an optimum recovery level, and (iii) produce sufficient concentrates for a testing campaign in Europe and Japan, in order to ascertain the suitability of Guelbs sinter feed for the blast furnace (SAR, June 1979, para. 5.06). Subsequently, commercial tests confirmed that the Guelbs material was a prime sinter feed, which was to permit a reduction of sinter costs when mixed with well-known European and Japanese base feeds. 21. Miniag at El Rhein, as in the existing Kedia operations, was to be done in open pits using 85 to 120 ton dump trucks for dispensing the ore into a primary crusher. Mining operations at Oum Arwagen (Phase 2) were to be basically identical. -5- Crushing and Bonification 22. Primary crushing was planned to be done at the El Rhein crushing yard to provide greater flexibility of operation. Ore from the Oum Arwagen mine was to be brought to the El Rhein crusher. Concentrates from the El Rhein plant were to be transported to F'Derik on a new 45 km railway line, linking the new operation to the existing rail system. 23. SNIM's consultants (SOCOMINE, a french engineering firm) first proposed dry pre-concentration at the El Rhein plant, to be followed by wet concentration near Nouadhibou (Basic Technical Report of April 1976). After trials (see para. 20 above), they proposed a new flow sheet (TSR on "Tout A sec au Rhein") which eventually involved a finer grinding of the preconcentrates (to 400 microns) at the mine, and eliminated the proposed bonif' ation at the port. This alternative was supposed to result in: (i) savings i , cransport; (ii) having all treatment facilities at one location for better control; and (iii) removal of difficulties connected with treating concentrates at the port with sea-water (contamination, corrosion). 24. The decision to use the all dry, low-intiensity, magnetic separation process was based on (i) what was then believed a general scarcity of water in the projct area; (ii) because of the assumed simplicity of the process; and (iii) the marketing advantage of sinter feed vis-&-vis the pellet feed that would have resulted from the wet process. 25. The dry process flow sheet to be used was relatively simple. Briefly stated, the crude ore is first crushed to 250 mm size and stockpiled. It is then reclaimed and conveyed to the bonification plant proper which is essentially made up of two identical and parallel process lines (line A and B), plus a third one (line C) for secondary ore treatment. Each main process line (A and B) consists of a large aerofalls grinding mill (the name is derived from the Canadian Company "Aerofall Mills") and a classification system. The ore is then screened and subjected to dry magnetic separation. All processes are linked together by belt conveyor systems. 26. A smaller size third line (C) is used to process, either the stockpiled grinding mills overflow, in much the same manner as done with the two main process lines; or again for secondary processing of the fine oxidized ores (as opposed to magnetic concentrated ores). In the latter case, the ore is first ground to 400 micron size in a dry ball mill, inclosed with circuit screens, and then concent;-ated on dry magnetic separators, again arranged in two stages of concentration. Thus two products are obtained: magnetic concentrates and oxidized concentrates. 27. Finally, the concentrate products are conveyed to stockpiles, from when they are reclaimed to be loaded in ore cars, for transportation to the seaport; whereas tailings are put in stockpiles near the mining site. 28. The solution adopted of grinding the oxidized fine ore to 400 microns by low-intensity magnetic separation was tested at the Zouerate pilot plant on 40 tons of feed. The results were mediocre, and no fully satisfactory solution was found regarding the need to upgrade ore to the necessary Fe content. -6- 29. A consultant hired by the Bank in 1976 during the pre-appraisal process had reservations on the efficacy of the adopted grinding and crushing process to 400 microns. This consultant found that "although the method using sea water provokes certain reservations, it is obvious that the matter of upgrading oxidized middling has not been fully resolved and would, therefore, require additional correlative test work." (Dolber and Company, geological and metallurgical consultants, New York, October 1976). 30. The consultants also noted that the sampling used of 460 tons of El Rhein's ore for concentration tests carried out in the SNIM's laboratory at Zouerate and abroad was "not qualitatively representative of the core assemblage in the deposits .... The principal thrust of their tests has been on dry, low- intensity, magnetic separation, because it is a relatively easy process and because of the scarcity of water." 31. The consultants generally concluded that "the TSR plan does not appear to have been as thoroughly thought as the earlier plan (wet concentration at Nouadhibou), and one cannot state unequivocally that it is feasible and desirable without further technical data." The processing of the project went ahead in spite of these reservations. Water 32. In 1979, proven water reserves could cover the Guelbs operation's requirements for a period of 30 years. While the industrial use of water was noZ to be appreciably raised, the increase in manpower meant a higher overall water consumption. Additional reserves were thus to be developed thereafter and a detailed drilling program was to be initiEted. Environment 33. In order to guard against generation of dust at the crusher station, whenever the water spraying system was out of order, a covenant in the Loan Agreement was introduced to the effect that COMINOR (SNIM's primary operating unit) would continue to operate its installations with due regard to ecological and environmental safety standards, and would carry out a long-term study on the effects of dust inhalation by the workers in order to determine whether adequate protection existed. If needed, suitable equipment was to be installed to protict the workers. Power 34. Since the generation station at Zouerate was worn out and could not be expected to supply adequate reliable power, E new power station was required. The bulk of the new power was to be ccnsumed Ly the grinding operation and the bonification plant. 35. A new road linking Zouerate with El Rhein was to be constructed during Phase I. The access road from El Rhein to Oum Arwagen was not to be constructed until Phase II. -7- Housing 36. An additional 528 housing units were to be built in Zouerate, in order to accommodate the additional manpower needed by the project. Railways 37. During Phase I of the project, four main locomotives as well as 293 X 100 ton ore wagons were to be purchased to handle transport requirements which were to increase upwards of 14 million tons of are per year. Additional equipment was to be acquired during Phase II and a new line built between El Rhein and Oum Arwagen. Port Handling 38. The main acquisition was to be a new shiploader with a loading capacity of 7,500 tph to replace the old one rated at 5,200 tph. Project Organization 39. The responsibility foi overall design, construction supervision, start- up, and commissioning of the project, and to assist SNIM during the first years of operation was assigned to SOCOMINE, a newly formed French consulting firm, consisting of personnel of the ex-MIFERMA's Technical Department, who had carried out the technical study of the Guelbs Project. 40. In view of SOCOMINE's small manpower of only 18 professionals and the large size of the project, the consultants and SNIM were led to seek assistance ,rom various other well-known French consulting groups, namely, SOFRESID, SGTE (Socidtd Gdndrale de Techniques et d'Etudes), and FCB (Fives, Cail, Babcock). 41. The key role assigned to SOCOMINE was justified as follows, "although it might appear that SOCOMINE's manpower is small, the fact that SOCOMINE's senior management were all former MIFERMA managers, who were involved in both the implementation and the subsequent operations of this company, reassured the Bank that SOCOMINE will be up to the task" (SAR, para. 7.02). However, while the consultant's staff was indeed involved in both implementation and subsequent operation of the Kedia mine, which did not involve any bonification dry process, it is difficult to see the applicability of this experience to the Guelbs project, which precisely involved a new and little tested technology. 42. This untested technology problem was, however, considered at the time of appraisal and it was decided to refer issues needing highly technical know-how to an expert committee (the "Technical Committee"), which was to act as a periodic review board. The Committee consisted of a senior engineer of SOCOMINE, represent ives of the French Steel Research Institute (IRSID), Mineral Services Incorporated (MSI , and an independent consultant. All had been closely involved in the feasibility study of the Guelbs project and in organizing the pilot plant at Zouerate (see para. 20 above). As the most critical equipment items in the project were the grinding plant (Aerofall Mills), detailed engineering was supposed to be checked at regular intervals and be subject to approval by MSI -8- and/or the independent consultant, as they had both been intimately involved in the original design of the system for Aerofall Mills in Canada. b) Proic.;t Cost and Finar.zing 43. Total financing requirements -- including working capital, interest during construction and contingencies -- were expected to be about US$500 million equivalent of which US$450 million (i.e., 90%) was to be in foreign exchange. The financing plan for Phase I envisioned US$338 million of long-term loans (i.e., 66% of total financing), US$120 million of new equity to be provided by five new shareholders (the Arab Mining Co. (ARMICO); the Islamic Development Bank, the Kuwait Foreign Trading, Contracting, and Investment Co. (KFTCIC); and the governments of Morocco and Iraq]. in addition to US$43 million from SNIM's internal cash generation. 44. The Bank loan of US$60 million (Loan 1747-MAU of December 14, 1979) was to finance the ore handling equipment, miscellaneous items in the bonification plant, the ore wagons and the civil works contracts, as well as interest during construction on the Bank loan itself. 45. Besides the Bank loan (which financed 12% of the project expected total cost), nine financial institutions provided US$278 million in long term loans: Saudi Fund, French Caisse Centrale, Kuwait Fund, Arab Fund for Economic and Social Development (AFESD), European Investment Bank (EIB), Abu Dhabi Fund, Japan OECF, African Development Bank and OPEC Special Fund.-! 46. In view of the financial risk resulting from Mauritania's limited capacity to service such large loans in the event of difficulties, a direct debt service payment mechanism was set up, to be built up and replenished by the Company's receipts of its ore exports. This was to assure automatic repayment of debt for a period of at least 6-12 months, thereby providing the lenders with an opportunity to seek a solution should serious problems arise. 47. A Set-Aside Agreement was put in place according to which all iron sales proceeds were to be placed in a trust account in London to accumulate the necessary amounts for debt service payments for six months and also maintain a permanent cushion equivalent to another six-month period. 48. Out of the remaining sales proceeds, certain minimum balances were to be transferred monthly to a SNIM Operational Account abroad to cover SNIM's foreign exchange operations as approved by the Central Bank. This was to ensure that SNIM had sufficient foreign exchange to meet its operational requirements. The remaining funds were then to be repatriated to Mauritania and credited to SNIM's account at the Central Bank. Clearly such arrangements provided lenders with special safeguards, in addition to the Government guarantee repayment of the 1 It was anticipated that Phase II of the project was to be financed by US$250 million in new loans, the balance (US$187 million) being derived from SNIM's accumulated cash surplus. Loans were assumed to be provided by international institutions (US$100 million) and suppliers credits (US$150 million). -9- Bank loan. Lenders overwhelmingly considered the Guelbs project as an "enclave" operation to be financed on relatively hard terms (except for Arab funds) and guaranteed by special Set-Aside and Security Agreements which provided them with an unconditional global assignment of all future proceeds from the sale of iron ore. c) Sales and Profitability 49. The sales volumes for both Kedia and Guelbs ore were assumed equal to production and were expected to evolve as follows: Iron - Ore Sales (million tons) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1994 (actual) Kedia 6.5 9.3 10.0 10.7 10.7 9.3 7.8 7.8 6.8 5.8 5.8 2.6 0.5 1.5 1.1 Guelba - - - - - 3.1 6.2 6.2 6.2 6.2 6.2 9.4 12.5 12.5 11.9 Total 6.5 9.3 10.0 10.7 10.7 12.4 14.0 14.0 13.0 12.0 12.0 12.0 13.0 14.0 13.0 50. These projections were devised on the basis of (i) existing long term contracts; (ii) letters of intent already signed; (iii) the favorable outcome of the Guelbs concentrates testing campaign executed by the existing consumers; and (iv) a Bank's market analysis which indicated that the supply/demand situation would be fairly balanced through the mid-1980s with the possibility of oversupply developing towards the end of the period 1985-1990, when new income investments, such as Carelas in Brazil, would start production. However, the Guelbs project was not assumed to experience substantial marketing problems, since the increment in iron ore supply which the project was to introduce on to the world market was rather small ( 0-4 million tpy, or less than 1% of total net exports). 51. These projections reflected a cyclical pattern comparable, to that experienced historically in the international ore market, and included a slow improvement up to 1980 (assuming a gradual recovery of the steel industry starting in the late seventies), while a simultaneous income in SNIM's production capacity as Phase I started up. This was to be followed by a high level of sales between 1983-1985 when world market demand and supply were expected to be in balance. After 1985, however, where many new iron mines might possibly have come on stream, another low cycle in the market was projected with reduced sales as a consequence. 52. The prices of iron ore were based on the Bank's world market projections, resulting in a 1977 FOB Nouadhibou price of US$15.20/ton for Kedia ore, and a projected base price of US$16.90/ton Guelbs ore (all in constant 1977 terms). - 10 - Average ore sales prices (US $/ton FOB Nouadhibou) 121 12Z2 19.1 194 1986 1982 1921 Constant 1977 Guelbs - - 17.2 16.9 14.5 16.9 Kedia 15.2 11.8 15.2 15.5 15.2 13.1 15.2 Current terms Guelbs - - - 28.2 31.6 33.2 44.4 Kedia 15.2 13.8 21.7 25.4 28.5 39.9 39.9 53. Based mainly on expected sales and ore prices, the incremental financial rate of return for the combined Phase I and II was a low 5.4% before payment of royalties to the Government (5% to 7% on FOB value of export) and 1.39% after royalties. It was assumed that in 1986, at full production of Phase I the project break even point for COMINOR would be about 66% of production capacity. It is to be noted that financial projections were made assuming that Phase I and Phase II would be carried out, implying that financial profitability would not be achieved with Phase I alone being implemented. 54. Despite projected low financial profitability, the overwhelming concern of the main shareholder, the Mauritanian Government, was the supposedly high economic benefit of the project, exter.-ing the life of the mine, and thus developing in the long term the major productive sector of the country's economy and the only sustainable activity in the northern part of Mauritania. This concern was also widely shared by some Bank staff (who insisted on the lack of alternative large viable projects in Mauritania) and by other lenders, in particular the Arab countries. 55. The economic rate of return was estimated at 11.8% which was relatively low but still higher than the projected 5.4% financial rate of return. This was due to the exclusion from the calculation of the important indirect taxes paid by SNIM and the high labor content of the operating costs (labor cost were shadow priced at 30% for laborers and 50% for foremen). In addition, the benefit streams were amended to take into account the cost of closing the mine, the town site of Zouerate, the communities along the railway line, and the Nouadhibou facilities, should the project not have been implemented. Few alternative means of employment or subsistence were available in Nouakchott, and none existed in the desert. Once the existing Kedia deposit was exhausted, all mining would have come to a halt and most of the population then estimated at 40,000 would have had to be resettled in other parts of the country (near Nouakchott or along the Senegal river). This resettlement cost was estimated at a minimum of US$2,000 per person and was included in the benefits of the Guelbs project. Without such an inclusion the economic rate of return would have obviously been lower. - 11 - (2) The SNIM Rehabilitation Project 56. After the 1979 Loan was approved, the Guelbs Iron Ore project had several serious problems, (i) a two-year commissioning delay (March 1985 instead of January 1983); (ii) SNIM sales in 1978-1983 decreased to about 7-8 million tons per year from the 10-12 million estimated at appraisal, mostly because of a depressed iron ore market; iii) iron ore prices in real terms were 15% below appraisal estimates; and (iv) unusual and severe technical difficulties occurred and productivity decreased. While these problems affected SNIM's financial position seriously, they were further compounded starting in 1985 by the start of the repayment of the US$338 million Guelbs loans. 57. After consultation with its lenders, including the Bank, SNIM embarked in 1984 on a comprehensive Rehabilitation Program consisting of a series of measures aimed at improvements in management and reductions in production costs. 58. In 1985, a third Bank loan, amounting to US$20 million, was made to SNIM to finance part of a Rehabilitation Project which itself was a part of a Global Rehabilitation Program for the company, designed to achieve productivity improvements as well as major modification. of the Guelbs plant. The Arab co- lenders agreed to make available unused portion of their Guelbs loans and reschedule some loan repayments. The CCCE made a new loan available for technical assistance for the Guelbs start-up. Total lending was US$55.8 million in addition to US$20 million contributed by the Bank. SNIM provided US$36.4 million in equity through cash generation. Total project cost was thus expected to be US$92.2 million. 59. Further financial assistance for the Rehabilitation Program was provided in 1987 for a total amount of US$61.6 million through loans from the ADB, the EC, the EIB, the Kuwait Fund and the CCCE. The Bank did not participate. Total financing for 1985-1987 thus amounted to US$154 million in addition to the US$500 million Guelbs project, giving an impressive total cost of US$651 million. 60. The Rehabilitation Program (of which the Bank-financed Rehabilitation Project was part) aimed at improvements and costs reductions in the following areas: (i) management and organization; (ii) operations; (iii) procurement and investments; (iv) personnel; (v) overhead cost; and (vi) cash management. (i) Management and Organization 61. As from 1986, a detailed five-year production and investment plan was to be prepared and updated annually. New systems for cost accounting, reporting and budget control were also to be introduced in 1986. (ii) OiperAtions 62. SNIM was to implement a program for equipment overhauls of installations in mining, port and railways, and control of spare parts use. SNIM was also to make arrangements for improvements of severe technical problems, in particular with the control of dust from the totally dry process. Finally, SNIM - 12 - agreed to achieve a production from the Guelbs plant of no less than one million tons over three consecutive months. (iii) Procurement and Investments 63. SNIM agreed not to invest more than US$5.5 million per year without approval of the Bank. Investments were restricted to the absolute minimum. (iv) Personnel 64. SNIM agreed to reduce personnel and overtime in order to reduce labor costs by about 3.5% per year in real terms. In the period 1985-1990, the projected savings in labor costs were to amount to US$23 million. (v) Overhead Cost 65. SNIM was to further reduce costs in areas such as its offices abroad and in Nouadhibou, non-iron and research activities, transport, and general training. (vi) Cash Management 66. The day-to-day management of cash was to be improved so as to reduce the short-term debt. Bank overdrafts were to be reduced by at least US$5 million by mid-1987. 67. The December 1985 US$92.2 million Rehabilication Project was part of the Rehabilitation Program. SNIM's internal cash generation, funds from co- lenders and the US$20 million Bank loan were to finance the foreign exchange cost of necessary items for the following purposes: * Replacement of Equipment (USS27.9 million). Some heavily worn railway track sections, locomotive and locomotive engines were to be replaced to ensure adequate transport capacity. Worn and obsolete mining equipment was also to be replaced. The Bank loan was not to be used to finance these items. * Major Overhauls of Equipment (US$33-4 million). Major fixed installations were to be overhauled to eliminate production bottlenecks. Spare parts and supplies (e.g. tires for mining trucks, conveying parts and wearing parts) were to be purchased. * Tools and Control Equipment (USS5.1 million). Sufficient and improved tools and instrumentation were to be provided for workshops and for major field equipment to improve efficiency and control. * Training and Technical Assistance (USS9 million). Consultant services were to be obtained for the implementation of a new management information system, mine planning and mine operations, railway maintenance, spare parts management and start-up of the Guelbs plant. About 24 expatriates were to continue to be employed to train Mauritanians in the areas of operations, maintenance and supervision. - 13 - 68. The estimated total project cost in mid-1985 prices (net of taxes but including contingen:ies) was estimated at US$92.2 million of which US$75.4 million was in foreign exchange. As indicated in para. 58, the Bank loan was supplemented by US$55.8 million in loans from SNIM's previous lenders for the Guelbs Project, whereas the Kuwait/Arab/Saudi/Abu Dhabi Funds and OECF made available unused portions of their Guelbs loans under the same favorable terms. The CCCE made a new loan of US$3.0 million. All loans were included in the existing Trust and Set-aside agreements for the Guelbs project. The allocation of the Bank loan by category included US$16.7 million for spare parts and supplies, US$1.3 million for the tools and control equipment, and US$2.0 million for training and technical assistance. Local costs (mainly labor) of US$16.8 million were to be financed by SNIM. 69. Financial proiections for SNIM were based on the assumption that the Kedia mines would be progressively depleted and that Guelbs production would rise so that SNIM's production/sales v-lume would stabilize at the 9.5-million-ton level reached in 1984 during 1985-1995. Production/Sales Forecast (million tons) 1984 1985 1986 1987 1988 1989 1990 1995 (actual) Kedia Rich Ore 6.0 5.3 4.0 3.0 2.5 2.5 2.5 2.5 Siliceous Ore 3 2.1 2 2.0 2.0 2.0 1. 1.5 Subtotal 9.5 8.0 6.5 5.0 4.5 4.5 4.0 4.0 Guelbs - 1.. ,. 45 0 5.0 55 55 TOTAL 9.5 9.5 9.5 9.5 9.5 9.5 9.5 9.5 70. This new forecast was much more conservative than the 1979 Bank forecast which had estimated sales at 14.0 million tons in 1984/85 (of which 7.8 million tons for Kedia ore and 6.2 million tons for Guelbs)(para. 49). Similarly, Kedia output had been assumed in the old forecast to be 0.5 million ton only in 1990, while Guelbs was expected to be producing 12.5 million tons, giving a total of 13 million tons against a revised forecast of 9.5 million tons. Finally, whereas the earlier forecast was a total output of 13 million tons (of which 11.9 million tons for Guelbs and 1.1 million tons for Kedia) in 1994, the new forecast has revised output downward to 9.5 million tons (of which 5.5 million tons only for Guelbs). 71. The drastic change in the forecast for Kedia ore output was partly linked to new estimates regarding the economic life of the mines. In 1978, reserves of high-grade ore (64% Fe) were estimated at 95 million tons, but after - 14 - eight years of exploitation (60 million tons were produced in 1978-1985) reserves were reassessed at about 30 million tons. Assuming in the future an average yearly extraction of about 3 million tons of high-grade ore (average 1985-1995), the reserves were assumed to last until about 1995. 72. There were larger reserves of siliceous, low grade ore (50-55% Fe) at Kedia, but siliceous ore was mined as a by-product of high-grade ore: it formed part of the overburden which had to be removed to get access to the rich ore. Mining of the low-priced siliceous ore (market restrictions for siliceous ore were expected to persist in the future) on its own, without the benefit of coproduction with high-grade ore, was judged to be probably not economical. Thus, the limiting factor for the expected life of the Kedia mines was the availability of the high-grade ore until 1995. 73. As the Kedia mines were being depleted, an increasing amount of Guelbs ore was to be mined (starting production in 1985 only, i.e. two years later than expected in the 1979 forecast). At the time of the Guelbs Project appraisal, proven raw ore reserves at El Rhein and Oum Arwagen had been estimated at 350 and 101 million tons respectively. However, the unfavorable development of iron prices necessitated a reassessment of economical reserves and it was estimated that only about 50-60% of the reserves may have been economically recoverable at 1985 ore prices. Since about 2.3 tons of raw ore were needed to produce one ton of product (after upgrading through magnetic separation), the economical reserves were to allow an operation of the Guelbs plant (6 million tpy capacity) over a period of 18 years. Most importantly, under the assumption of continuing low iron ore prices, there was to be no reserve base for a Phase II Guelbs plant, (SAR, November 12, 1985, para. 2.07). It is to be recalled that the Guelbs Project had been originally conceived with two phases in order to reach a 14- million-ton production level and ensure the Project's financial viability. 74. Despite projected savings in labor cost (para 64) and in other operating costs, SNIM's operating costs were expected to substantially increase due to increasing production of the Guelbs (from 1.5 million tons in 1985 to 5.5 million tons in 1990) that required the costly bonification dry process. This process (including crushing and handling) was to cost US$4.8 per ton of product in addition to US$2.6 per ton for mining costs. This total of US$7.4 compared with US$3.2 only in Kedia (at 1984 prices). As a result, operating profit as percent of sales was expected to decrease from 19.1% in 1985 to an average of 10.1% in 1986-1990. Moreover, assuming that SNIM would continue to pay a royalty to the Mauritania Government amounting to 10% of sales of iron ore, net income as a percent of sales was to be negative in 1986-1989 and become positive in 1990 only with 4.2%. It was, however, indicated that in view of the importance of SNIM to the country, the Government may have to agree to reduce royalty payments which were high compared to that paid by SNIM's competitors (SAR, November 1985, para. 7.05). 75. The financial rate of return for the Project itself was calculated on an incremental basis in real terms. In addition to the assumptions concerning labor and other costs (mainly materials) savings, it was assumed that the Rehabilitation Project could be credited with a level of production of one million tpy higher than in the without-project case (SAR, November 1985, para. 7.12). - 15 - 76. Based on these assumptions, the anticipated rate of return was 21%. The rate of return of the Guelbs Project, given actual project cost, current forecasts of iron ore prices (see para. 78) and updated estimates of production costs was recalculated at -3% as compared with 5.4% forecast in 1979. The financial rate of return of the Guelbs Project combined with the Rehabilitation Project was estimated at 1%. 77. The economic rate of return was calculated with no shadow pricing applied for the foreign exchange since the Ouguiya was a free-floating currency. All taxes and royalties were excluded. The labor wage was judged to reflect appropriately the economic cost of labor consisting of foregone production in employment outside SNIM plus the cost of training necessary for employment with SNIM. The economic rate of return of the project calculated under these assumptions was 25%, slightly higher than the financial rate of return. 78. The above rates of return assumed an expected reduction of iron ore prices by 1995 by 7% in real terms from its 1984 level. This was based on the fact that there was in the mid-1980, a substantial excess world iron production capacity and a severe contraction of the international iron ore market in the face of the steel market collapse. While the Guelbs Project had been appraised on the assumption that the price nf Guelbs ore (FOB Nouadhibou) would be US$28.2/ton in 1984 and US$33.2 in 1989, the Rehabilitation Project forecast the price of Guelbs ore to be US$22.5 per ton in 1989 (at current prices), i.e., one- third lower. In 1984 the actual price was US$17.9 per ton, i.e., 37% below the price which had been forecast in 1979 for that year. III. IMPLEMENTATION 79. While the implementation of the Guelbs Iron Ore Project was affected by a multitude of serious technical problems, the subsequent SNIM Rehabilitation Project was implemented mostly without difficulty. A. GUELBS IRON ORE PROJECT 80. The Project was to be implemented between end-1979 and 1983 with the core bonification plant to be completed by March 1982. However, the latter was completed in May 1984, i.e., 24 months later than anticipated. Erection of the power plant was also delayed by 20 months, while mining and railroad components were implemented ahead of schedule (six and two months respectively). 81. Although it was estimated at appraisal that the plant would be in full production by the end of 1983, or one year after completion of its construction, in September 1985 the plant had not yet, in fact, progressed to the start-up period (cf. Report by M.A. Hanna Company, Cleveland, Ohio, September 27, 1985). Only 760,000 tons of concentrate had been produced and no more than one million tons could be expected to be produced for the whole year of 1985 as compared to the 6.2 million tons originally estimated for 1985 (the 1979 forecast was that production would reach 3.1 million tons in 1983, increase to 6.2 million tons in 1984-1988 and 12.5 million tons in 1990 and after). Actual production rose to 1.4 million tons in 1986-1987 and to about 2 million in 1988-1990, but never reached the expected levels. Six years after completion of construction, the - 16 - plant has attained only one-third of its initially assumed capacity. The original estimate of achieving full production only one year after completion of construction was a grossly over optimistic assumption, not only in retrospect, but even at appraisal (PCR, para. 5.06). (1) Implementation Delays 82. The delays, which occurred during project implementation, are attributable partly to suppliers and partly to the Engineers' lack of experience in executing projects of this complexity (PCR, para. 4.05). There were three main types of delays: (a) delays in loan effectivoness, (b) delays in procurement, and (c) delays in construction. a) Delays in Effectiveness 83, The process of project preparation had already been a lengthy one. A first identification mission in January 1977 was followed by an appraisal mission in October 1977. The complexity of the project required a total of 12 missions up until 1979, because: (i) SNIM had to be reconstituted and made into a viable borrower (principally through the divestiture of its highly unprofitable copper operations); (ii) four lenders' meetings had to be convened in order to develop a viable financing plan; (iii) meetings with banks and iron ore consumers were necessary in order to structure an acceptable security arrangement for the parties involved; and (iv) extensive working sessions had to be held with SOCOMINE, the project consultant, in order to ensure proper project management. 84. The Board finally approved the loan in July 1979 and loan signature took place in December 1979, three years after the first identification mission. Concerns were expressed by some Board members on the marginal character of the project and the low financial rate of return (5.4% only even after completion of Phase II). Bank's management pointed out that iron ore production was very important for Mauritania and that the Bank was hardly in a position to reject the Guelbs project which was also financed by Arab and French lenders. 85. It took one year after loan signature for the loan to become effective. This was mainly due to a legal issue raised by the Mauritanian Government regarding the proposed Trust and Set-Aside Agreements. 86. Further delays were also experienced as a result of cross-effectiveness requirements that, at least, 80% of total external financing should be ratified for the Loan Agreement to be effective. Due to slow ratification procedures of some lenders, effectiveness only occurred in December 31, 1980. Effectiveness delays did, however, only moderately affect project execution (by about four months) since a few early equipment supply contracts could not become effective, while downpayments for most critical contracts were made by SNIM from own cash generated (PCR, para. 5.01). b) Delays due to Procurement Problems 87. Whereas by July 1981, according to appraisal expectations, all contracts were expected to be signed, in reality 56% had been signed, 24% were at the stage of negotiation or preparation, and 20% under preparation or - 17 - outstanding. There were also delays of six to twelve months between selection of suppliers and signature of contracts. Delays in key contracts showed, in 1981, the following pattern: Number of Average Delay Causes of Delays Critical Contracts (months) Delays in Effectiveness of Contracts 18 4 Delays in Selection of Suppliers 3 2 Preparation of Bid Investments 3 7 Other Causes 1 TOTAL 25 Source: World Bank/Kuwait Fund Joint Supervision Report, November 1981. 88. Delays in contract award were due to protracted negotiations, missing engineering drawings and contract documents, in disbursement of down payments and signing of approved contracts. 89. Some bids for the Bank financed contract for industrial civil works were rejected because prices were considered too high in the Engineers' opinion. The re-bidding resulted in a three-month delay and with new bids at equal or higher prices. It was, therefore, a wrong decision to call for re-bidding (PCR, para. 5.02) since the only significant result was to delay project implementation. c) Delays in Construction 90. Trials runs of lines A and B of the bonification plant took place in July/October 1984, while it had been originally anticipated that they would take place in January 1983. Construction delays were mainly caused by several interrelated factors (some of them were mentioned above), namely: (i) delays in effectiveness; (ii) delays in disbursements; (iii) delay by suppliers in meeting delivery target dates; (iv) supply of defective equipments which had to be modified or re-manufactured; (v) inadequate experience of the Engineers to handle such complex project; and (vi) frequent managements changes by SNIM. 91. Problems also occurred due to: (i) coordination problems between the engineers and SNIM's management due to inaccuracies in consultants' plans which failed to serve as a tool for inter-departmental coordination (Supervision Report, September 14, 1981); and (ii) delays in lenders' approval and payment procedures which often took too long and delays in providing progress reports on status of implementation. - 18 - (2) Start-un Technical Problems 92. By end-1985 there was a considerable amount of dust, mechanical and process problems being experienced with much of the plant equipment. In particular, since the bonification plant was still in the "transitional period" - - that is the period between the preliminary and final acceptance of the equipment -- and was, therefore, not yet legally turned over to SNIM, the Company felt it had to arrive at mutually agreeable arrangements with the equipment suppliers so as to provide them with the required plant time to perform necessary modifications and deficiency corrections. 93. In September 1985, "compliance certificates" (merely confirming SNIM's agreement to the effect that the equipment had been supplied and erected in conformity with plans and specifications) had been iasued for all major equipment. However, "preliminary acceptance certificates" (issued after the equipment had been run for a given period of time and tests had demonstrated its mechanical and electrical reliability) had not been issued for key items such as conveyors, grinding mills, magnetic separators, stackers and wheel reclaimers, because of serious deficiencies. Of course, no final acceptance (given only after it is demonstrated that the eouipment is operating at lated capacity) had been given either, since the plant or components of it had not been able to operate at plant capacity. 94. Serious technical problems whiclh delayed the plant start-up and have still not been fully overcome were the dut problem, the eauipment deficiencies, the grinding mill capacity problem and the separation process. a) Dust Problem 95. Notwithstanding the positive findings of all the earlier studies and expert assessment by ILO and SNIM and its engineering units, the volume of dust has been much greater than expected. In fact, extremely dusty conditions prevail throughout the whole bonification plant. The entire Guelbs complex is eften enveloped in a huge cloud of dust that can be seen from 10 to 15 kilometers away. For this sort of thing to be happening, when the plant is running at only one- third of its rated capacity, project designers, equipment suppliers, and possibly the owner, must have completely overlooked or grossly underestimated the magnitude of the dust problems associated with the handling and processing of very dry ores and thus the need for providing a minimum amount of dust control equipment. 96. The conception of the closed air circuit system, which carries the dust, was ill-conceived due to a gross underestimation of the problem, and to an ill-advised early decision to eliminate the dust filter mechanism at the Pilot Plant for reasons of cost. b) Equigment Deficiencies 97. Together, with the dust, they were the major causes of the plant start- up delays. - 19 - (i) Crusher. It experienced stalling problems due to overheating of the lubricant and the starter motor's capacity was inadequate to operate when the crusher was underload. This was noi foreseen during the planning stage of the project. (ii) Conveyors. The conveyors failed to guide falling ore flows into the center of receiving belts properly, causing them to shift sideways and spilling. Also, the back sides of transfer points were too steep causing ore slides, rollbacks and spills, this generating even more dust. Part II of the PCR (para. 18) stated: "Although no one can doubt KRUPP's ability to supply efficient conveyors and the selection report faulted it only for its high price (which could have been lowered), the company was completely non-responsive and did the project a great deal of harm .... The conveyor system was poorly designed and ill-suited to the plant." (iii) The automatic sampler. The sampling mechanism and bucket was too light for the size of crushed ore handled and therefore tended to constantly break down. (iv) Reclaimers. The two crushed ore wheel reclaimers were too light for the size of material they had to handle. The stationary retaining side plate which prevents the wheel buckets from emptying out before reaching the belt transfer point was too light. It bent out of shape too easily, with the results that pieces of rock lodged themselves between it and the wheel ribs, tearing and breaking them out. (v) Liners. Liners in parts of the grinding mills, classification systems and fans wore out at a much faster rate than anticipated. (vi) Triper car rails. They were already worn out in 1985 and harder type of rails had to be replaced. (vii) Magnetic separator drives. The magnetic separator drive chains and sprockets were not properly protected from the dust and, wearing out at an exceedingly fast rate, had to be replaced with harder ones. (viii) Exhaust fans were unprotected. (ix) Ventilation circuits and shielding suffered heavy wear. (x) Spare parts requirements were underestimated when contracts were signed. c) Grinding Mills 98. They did not succeed in reaching their rated capacity. One of the problems may have been due to using either too large a size of grinding balls (125 mm balls) or that the size of the ball charge was wrong. They were major errors in designing the plant's process (PCR, para. 4.04). 99. Dry grinding mills similar to SNIM's exist only at Carrol Lake of International Iron Ore Company, Canada, where it also took a relatively long time to reach production. At the time when the Guelbs plant was designed, Carrol Lake - 20 - was still in the start-up phase. A full transfer of experience was, therefore, not possible. Extensive pilot testing and careful engineering were not able to avoid errors in designing the big grinding mills. d) Separation Proces1 100. While product specification and recovery rates for magnetic ore were as expected, processing the oxides in two separate stages proved difficult, whereas, on the strength of the pilot results, these products were seen as the "real meat" of the project. First-stage concentrates proved good but the second- stage products had to be sold as middlings (preconcentrates) due to their lower- than- expected Fe content. 101. Despite extensive testings, the sampling method adopted for the oxidized ores during the research phase appears to have been faulty. Guelbs' production targets are not being achieved nor are they likely to be achieved in the short-run until a solution is found to the treatment of oxidized ore which has been suspended due to low concentrate yields. B. SNIM REHABILITATION PROJECT 102. The Rehabilitation Project was implemented much more satisfactorily, although a number of problems faced during the Guelbs Project implementation phase have not yet been fully solved. 103. Equipment replacement and overhaul financed under the Project was initiated in January 1984 as expected and completed in December 1988, with a six- month delay. Training and technical assistance components did not suffer any delay between January 1984 and July 1988. Finally, managerial improvements provided under the Project were implemented ahead of schedule, i.e. by March 1987 instead of December 1987 as originally planned. 104. Modifications and replacements have been ongoing to the present, particularly in view of the fact that the 1985 loans for SNIM Rehabilitation were followed in 1987 by other loans, totalling US$50.8 million, from Atab and French lenders as well as by the EC, EIB, and ADB (the Bank did not participate). A US$40 million IDA credit approved in 1990 is being partly used (together with other resources provided by Arab, French and Japanese sources) to further help SNIM in completing the Guelbs technical and managerial action program. 105. With assistance of a manager/engineer, who played a key role in starting Carrol Lake in Canada (the only other plant comparable to Guelbs), an appropriate action program was again formulated in 1988, which is currently being implemented. The program includes steps to reduce dust leakages, modifications of the mill air circuit, better mill load control, as well as changes in material flows. In October 1989, SNIM also implemented wide ranging managerial changes. Output of the plant has increased from 1 million tons in 1985 to about 1.4 million in 1986-1987, and about 2 million tons in 1988-1990. However, higher levels of production do not seem likely to be achieved in the near future due to suspension of processing of oxidized ore and continued technical problems. - 21 - SNIM - STATUS OF IMPLEMENTATION OF PROGRAM OF MAJOR ACTIONS FOR GUELBS PLANT ACTION STATUS AND TARGETS 1. Strengthen Employ consultant for overall Started in November 1988, on- Management management and plant going, 2-3 interventions per maintenance year of 3-5 weeks duration. Employ consultant for Started In June 1988, all engineering of plan major design work completed, modification. assistance continued on an as- needed basis for supervision of construction. Employ consultant for Started in March 1989, training. combined with consultant services for overall management and plant maintenance. Assign new plant management. New plant manager was assigned October 1989. 2. Improve Plant Install major dust filters at Dismantling of redundant magnetic separation. filters from mill air circuit started January 1990, reassembly at magnetic separation to be completed May 1991. Change mill air circuit from orders for equipment have been partial recycling to 1002 placed, installation open. completed. Install electronic mill load Was installed in March 1989. control. Reinforce reclaimers for order has been placed, feeding crushed ore to mills modification has been completed on one and will start on second. improve transfer of ground ore order for equipment has been to magnetic separation. placed, modifica- tion to be completed in 1991. 3. Achieve Targetsa' Produce 2 Mt/a. Has been achieved in 1989. Produce 2.5 Mt/a (1990). Targets could not be achieved due to suspension of Produce 3.5 Dt/a (1991). processing of oxidized ore and continued technical problems. minimam production targets for period of executing the plant modifications. - 22 - IV. RESULS A. THE GUELBS IRON QRE-PROJECT 106. Iron ore production haa been much lower than anticipated at appraisal, i.e. one million tons only in 1985 (when Phase I of the Project was completed) instead of the expected 6.2 million tons. As explained above, design problems and severe implementation difficulties have prevented the Project to reach its production objectives. 107. The mine and infrastructure componer.ts of the Guelbs Project were completed without problems. The Guelbs mine is a modern, efficient and low cost operation, and SNIM has been able to achieve record petformance of its railway (PCR, para 6.03). 108. The problem lies with the Guelbs bonification plant. Dust and equipment deficiencies have been major factors in precluding the plant to produce more than two million tons in 1990, i.e. only 33% of the nominal capacity. (1) Plant Availabilitv. The initial underestimation of dust generation and its impact may for ever keep plant availability below the 1979 appraisal forecast, although current action program anO managerial changes may still increase availability above the low present 40% ratio. (2) Throughout Rate. Hourly throughput, i.e. tons produced per hour by the magnetite ore and the oxidized or. lines of the bonification plant, is reported to have reached 800 t/h against 895 t/h forecast at appraisal (i.e., 78%) for magnetite ore and 800 t/h against 1,000 t/h (i.e., 80%) for oxidized ore. The difference with appraisal estimates is mainly due to a wrong scaling-up factor from the pilot plant tests and to an underestimation of ore hardness. 109. Due to reduced availability and throughput rate, a derating of the nominal capacity of the Guelbs plant is necessary. Sore experts have 7redicted that the plant may finally have a production capacity of about four million tons per year of concentrate, instead of six million tons originally expected. 110. Phase II of the Project, which was imperative both in order to maintain SNIM output at 12-14 million tpy (due to Kedia mines declining production) and thur. to assure thu Company financial viability was abandoned several years ago due to the difficulties to reach even part of Phase I objectives. This was compensated by the fact that, contrary to earlier expectations, Kedia output was higher in 1990 than in 1980 as shown below. - 23 - Crushed Iron Ore Production (1980-1990) (million tons) 1980 1981 1982 1983 1984 1985 1986 198 1988 1989 1990 Kedia Actual 8.9 8.4 8.5 6.5 9.5 8.2 8.0 7.7 7.6 9.9 9.4 1979 Forecast 10.0 10.7 10.7 9.3 7.8 7.8 6.8 5.8 5.8 2.6 0.5 1985 Forecst - - - - 9.5 8.0 6.5 5.0 4.5 4.5 4.0 Guelba Actual - - - - 0.3 0.9 1.5 1.4 1.9 2.0 2.0 1979 Forecasty - - - 3.1 6.2 6.2 6.2 6.2 6.2 9.4 12.5 1985 Forecast - - - - - 1.5 3.0 4.5 5.0 5.0 5.5 ( Including Phase II starting in 1989.) 111. As a consequence of the production problems at Guelbs, SNIM started in 1986 an intensive exploration effort focusing on Kedia and on some high grade occurrences in the vicinity of Guelbs. Additional high grade ore reserves were discovered at Kedia, which proved to be at least 40 million tons (of which 20 million tons discovered in 1986). These reserves were larger than anticipated at appraisal and have enabled SNIM to maintain, and even go above, the production levels reached in 1980-81. Before the 1986 discovery, the 1985 appraisal of the Rehabilitation Project had revised downwards Kedia projected output for 1986-1988 due to unfavorable development of iron prices necessitating a reassessment of economical reserves. Such forecast proved too pessimistic after the 1986 discovery. This has meant that SNIM has been able to compensate the Guelbs Project failure to meet its objectives through increased production at Kedia, a complete reversal of what was originally anticipated. 112. Due the discovery of additional ore reserves, the transfer to exploitation of low grade ore at Guelbs lost its previous urgency. Furthermore, due to the much-lower-than-estimated iron ore prices, continuation of the high grade ore operations at Kedia at a high level of production and at a lower cost (since there is no bonification process used at Kedia), was needed for the Company's survival. Thus, the second phase of the Guelbs project could not, any more, be conceived as a viable operation given continuing low iron ore prices and was, therefore, postponed indefinitely. All this shcws that in retrospect the Guelbs Project would not have been necessary at te chosen time and a more - 24 - thorough evaluation of even minor indications of additional high grade ore reserves should have been conducted (PCR, para. 4.02).1i 113. As part of its intensive exploration campaign, SNIM also focussed on some known high grade occurrences in the vicinity of Guelbs. As a result, the M'haoudat deposit was delineated. The drilling carried out since July 1987 has indicated reserves of about 80 million tons (about 65% high grade and 35% low grade), which can be exploited without bonification, allowing a 5-6 million tpy operation over a period of about 15 years. Although relatively small, M'haoudat provides SNIM with a promising opportunity to keep the present level of sales after the Kedia deposits are depleted. The discovery of the M'haoudat deposit was made by following up on minor surface indications of high grade ore, which at the time of 1979 Project's appraisal were known but were interpreted as too small for commercial exploitation. Here, again, more serious exploration efforts should have been made at the time, rather than too hastily assume that there was no high grade ore to be found in the area. 114. SNIM also intensified exploration of additional water resources and such resources have been found. It was concluded too early that there was no water in the area, and the search had focussed only on a small 50 km. strip around Zouerate. Largely due to the so-called "scarcity of water," an untried dry process was selected for the bonification plant with the disastrous consequences described above. If more exploration efforts had been made for additional water resources, (such resources were effectively found later on), the project design could have been entirely different. 115. The recalculated financial return of the Guelbs project is negative (-10%) and can be compared to the already abnormally low 1979 appraisal estimate of 5.4%. As stated, such a negative result was due to the severe technical problems of the plant and resulting high production costs. Another key factor was the much-lower-than-estimated iron prices as shown below: Cf. footnote, para. 152. I Expertu from the European Investment Bank are of the opinion that a wet process, even if feasible, would have met a stiff opposition from ecologists. - 25 - Iron Ore Prices - FOB Nouadhibou (current prices in US$/metric ton) 1984 1985 198 1987 1988 1989 1990 Guelbs Actual 17.9 17.2 15.9 14.7 14.4 15.8 18.1 Projected 28.2 30.2 31.6 33.9 33.5 33.2 38.4 (1979) Projected - 18.2 19.4 20.3 21.4 22.5 23.7 (1985) Kedia Actual 15.9 17.1 16.2 15.6 13.0 19.6 22.7 Projected 25.4 27.1 28.5 30.0 30.2 29.9 34.5 (1979) Projected - 16.4 17.2 17.9 18.7 19.7 21.1 (1985) 116. The Bank's iron ore price forecasts have consistently been over optimistic as shown in the above table. Due to depressed steel demand and severe competition on world markets, world iron prices have not been very stable. For Guelbs, the average ore revenue per ton fluctuated from US$17.2 per ton in 1985 to US$14.4 in 1988 and US$18.1 in 1990 in current terms. Measured in constant 1977 terms, FOB Nouadhibou prices dropped 45% from 1977 to 1989 (US$16.9/ton to US$9.3/ton). Exchange rate fluctuations have amplified the impact on SNIM's financial situation. In constant terms, expressed in major European currencies, the 1989 iron price was about 40% lower than that of mid-1985. This had had significant impact on SNIM's finances, since about two thirds of SNIM's supplies come from Europe (representing about 40% of its total cash expenditures), while its revenues are in US dollars. 117. The recalculated economic rate of return was a negative 5% compared with 11.8% at appraisal. Similar assumptions than at appraisal were made regarding the use of shadow price for labor (35%) and the elimination of indirect taxes, representing about 3% of FOB export value. Similarly capital costs were reduced by 3% to take into account indirect taxes and shadow-priced labor. 118. In 1986-1990, the share of SNIM's revenues generated by sales of Guelbs iron ore has remained around 17% as a result of the slow increase or stagnation of Guelbs output. In contrast, about 70% of SNIM's direct costs originated in the Guelbs Project. Debt service payments for Guelbs have increased from 49.5% in 1986 to 94% in 1990. The totally unsatisfactory return of the Project led to non-compliance with the financial covenants under the Loan Agreement, e.g. SNIM debt/equity ratio which was not to exceed 1.5 rose to 1.7 in 1987 and 1989, while the current ratio, which was not to be less than 1.3, fluctuated between 0.6 and 0.8 in 1985-1989. - 26 - Contribution of Guelbs Iron Ore Proiect to SNIM's Financial Performance 1985 1986 1987 1988 1989 1990 Sales Volume 9.3 8.9 9.0 10.0 11.1 11.4 (million tons) Guelbs Contribution 8.6 18.0 14.4 19.0 18.9 17.5 (%) Revenues (US$ 160.5 145.5 137.7 132.5 211.8 257.3 million) Guelbs Contribution 8.6 17.5 13.9 20.7 15.7 17.1 (%) Direct Costs (US$ 23.4 19.4 24.6 20.8 17.8 22.3 million) Guelbs Contribution 71.4 66.0 73.9 70.4 70.4 67.9 (%) Debt Service (US$ 29.3 63.9 111.9 44.5 45.8 44.4 million) Guelbs Contribution 88.1 49.5 38.9 89.4 91.3 94.0 (%) 119. On the positive side, at its current production level of two million tons per year, all direct fixed and variable cash operating costs (before interest) are covered. In 1992, all financial charges related to the Guelbs debt would also be fully covered. 120. Guelbs Project total cost was ITS$414.3 million, i.e. 17% below appraisal estimates of US$500.7 million. Installed costs amounted to US$372.6 million, i.e. 15% lower than the appraisal estimate of US$139.3 million. The only large cost overrun was for training and preliminary work (US$45.3 million instead of US$15.0 million). However, if all cash operating losses, plant improvements and interest payments incurred, after project completion in March 1985 up to 1990 are considered, a cost of US$100-120 million would have to be added. The Projects' final cost, broadly defined, therefore is likely to be somewhat US$500 million (PCR, para 5.04). 121. The environmental impact of SNIM's mining operatiGns is low due to favorable geological and geographical conditions and the use of modern mining equipment. There is relatively little waste material from the mines, due to the low overburden/ore ratio. The material consists of solid inert rock and is deposited in a safe manner in rugged mountain valleys next to the mine, in an uninhabited desert area without rainfalls. The rejects of the totally dry Guelbs bonification process are equally inert and are stacked in low, safe benches with modern stacking equipme..t next to the plant. Dust generation within the Guelba plant is high, but the measures under way to improve dust problems should - 27 - hopefully reduce the dust load in the plant. The health record of the Guelbs staff is good (SAR, PESAP, Annex VIII, para. 50, May 30, 1990). B. THE SNIM REHABILITATION PROJECT 122. Results of this Project have been generally positive. The main objectives, which were to: (i) help SNIM to improve its management and reduce its production costs; (ii) assist with the necessary financial restructuring of SNIM by transforming part of its short-term debt/overdrafts into long-term deb.:, have been substantially achieved. 123. Equipment purchases and overhaul as well as training and technical assistance have taken place as expected. Total cost was US$87.1 million, i.e. 6% less than originally provided by various financing sources. The Bank loan was totally disbursed but loans by Arab Funds were not. 124. Major overhaul of fixed installations was completed and control over scheduled maintenance has been heightened. Measures were taken to improve spare parts management, including computerization of stocks and a reorganization of warehouse staff. Cost savings were achieved through consolidation of the procurement and purchasing departments, increase of technical staff and greater recourse to international competition of suppliers. Overhead costs were also substantially reduced. Apart from the transfer of SNIM's headquarters from Nouakchott to Nouadhibou in 1984, i.e. closer to operations, SNIM's steel mill, hotels, restaurants, clubs and supermarket were spun-off. The size of SNIM's Paris office was reduced. Three company airplanes were sold and SNIM is no longer providing free electricity to its staff (SAR, Public Enterprise Sector Adjustment Program, Annex 8, para. 11, May 30, 1990). 125. The work force was reduced from 5,913 in 1985 to 4,513 in 1989, i.e. by 24%. A drastic layoff took place in 1987 when staff was reduced by 1,446 people. Simultaneously, overtime utilization was curtailed and the number of expatriates cut from about 400 in 1980 to about 60 at present. Considerable emphasis has been given to training of staff. 126. SNIM's personnel costs fell from US$40.5 million in 1985 to US$36.6 million in 1989, i.e. a 10% reduction, while production per employee has increased from 1,545 tpy in 1985 to 2,638 tpy in 1989, an improvement of 71%. 127. Total cost savings achieved, so far, as a result of the rehabilitation program, are estimated in the order of US$11.4'million per year i.e. 35% higher than the appraisal estimate (PCR, para. 6.04). In 1990, average cost of goods (before depreciation and interest) decreased from US$12.6 per ton in 1985 to US$10.7 in 1990. SNIM's mining costs compared well with its major competitors, but its total average production costs are still somewhat higher because of the need for costly bonification of the Guelbs ore. 128. The overall financial situation of the Company has improved due to SNIM's cost savings measures, higher export sales and better iron ore prices in 1989 and 1990. In the two years, despite a steep increase in depreciation and financial charges for Guelbs, lower production costs (see para. 44 above) and increased sales (in 1990, the sales volume was 11.4 million tons and the average - 28 - revenue was US$18.2 per ton) combined to produce for the first time in years a subtotal US$26.7 million profit (after tax) (see Annex Table). In contrast, net losses had taken place in 1987 and 1988 when the most important causes had been increases in financial charges associated with excessively large debt (long- and short-term), as borne out by the low current assets/liabilities ratio (0.6 in 1987-1988) and a debt service ratio of 0.6 in 1987-1988. 129. A slight improvement in the world demand for iron ore has enabled SNIM to increase exports (from the Kedia mines since the Guelbs mine production stagnated) by 11.5% in 1990 compared to 1988. Current ore prices were also 33% higher in 1990 than in 1988. Moreover, substantial debt relief from various lenders has improved the current ratio and SNIM's net internal cash generation. 130. The various Security and Trust arrangements put in place for the original Guelbs Project in 1979 and which were extended to the loans provided for the Rehabilitation Project have assured that SNIM has always met its interest and repayment obligations. But to accomplish it, SNIM had to resort to overdraft financing and has been building up its accounts payable. SNIM has now been able to reduce some of its bank overdrafts and suppliers credits. In 1990, SNIM's current ratio improved to 1.4 after remaining at an unsatisfactory average level of 0.7 in 1987-1989. 131. The recalculated financial rate of return of the SNIM Rehabilitation Project has been calculated (PCR, Table 6) at 21%, or the same as at appraisal in 1985. The recalculated economic rate is 19%, only slightly lower than estimated at appraisal, i.e. 25% since substantial cost savings have been achieved as well as lower operational costs. 132. This satisfactory rates of return have partly compensated the poor results achieved by the Guelbs Project as shown below: Financial Return (%) Economic Return (%) SNIM Joint SNIM Joint Guelbs ehab Projs. Guelbs Rehab, Prois (1) (2) (1+2) (1) (2) (1+2) Appraisal 5.3 - - 11.8 - - (1979) Appraisal -3.0 21.0 1.0 2.0 25.0 5.0 (1985) Actual -10.0 21.0 -2.0 -5.0 19.0 0.0 133. Rates of return of the joint projects are still highly unsatisfactory. In order to achieve a satisfactory rate of return, iron ore prices would have to increase by almost 100% or the investment would have to be devalued to 10% of its actual cost (PCR, para. 6.06). - 29 - V. SUSTAINABILITY 134. The Guelbs Project annual production level has been about two million tons in the last three years due to major equipment overhaul and productivity incfeases achieved through the SNIM Rehabilitation Project, cost savings have boen achieved. The cash break even point (defined as the minimum production level at which the revenues from the sale of the Project's product cover all operating costs excluding interest and depreciation of the Guelbs mine and plant) has been reached in 1988 (PCR, para. 3.01). Revenues have exceeded direct costs in 1990 as shown below: Guelbs Revenues and Direct Costs (US$ million) 1985 1986 1987 1988 1989 1990 Revenues 17.6 24.2 20.8 27.2 33.0 22.4 Operating Cost 20.4 27.3 26.1 31.9 32.5 35.3 Cash Generation (2.3) (5.1) (5.3) (4.7) -- 2.1 Production Level (in 0.9 1.5 1.4 1.9 2.0 2.2 million tons) Operating Cost (US$/ton) 22.7 18.2 18.6 16.8 16.51 7.0 135. However, financial charges (i.e., depreciation and interest rates) are now very high. They amounted to US$47.2 million in 1989 and US$39.2 million in 1990. As a result, net losses amounted to US$47.2 million in 1989 and to US$37.1 million in 1990. Financial charges will progressively decline in the years to come with debt relief and repayment of earlier loans but will still be too large to enable the Guelbs project to make a profit even assuming that production, as a result of ongoing investments would reach 4 million tons by 1996. - 30 - Cuelbs Production. Costs. and Profitability 122 1992 2 122 1995 1996 200.0 Production 2.2 3.0 3.5 3.5 3.5 3.5 4.0 4.0 (millions tons) Iron Ore/ 17.0 16.5 16.0 15.5 15.1 14.6 14.6 14. prices constant 1990 prices (US$/ton) Revenues 37.4 49.6 56.1 54.4 52.7 51.0 58.5 59. (US$ million) Operating 35.3 44.5 50.3 50.3 50.3 50.3 56.0 56. Costs (US$ million) Financial 39.2 33.3 27.5 24.9 22.9 19.4 16.7 15. Charges (US$ million) Net Income (37.1) (28.2) (21.7) (20.8) (20.5) (18.7) (14.2) (12. (US$ million) 136. Although iron ore prices have increased in real terms in 1989-1990, most experts do not forecast a continuation of such increase but rather a slow decline during the present decade. Even arising a stability of iron ore prices in the 1990s, Guelb's net income would be still negative in year 2000. 137. Fortunately, prospects for SNIM as a whole seem to be brighter. SNIM is in a position to keep producing ore from the Kedia mines until the mid-1990s (given the declining ore reserves' situation). Preparations are now being made for the implemenL%tion of the high grade Mhaoudat ore deposits (80 million tons). The M'haoudat project cost is estimated at US$154 million and may reach full output by 1995. The 1990 IDA Credit for Public Enterprise Sector Adjustment Region together with loans from other lenders took into account the need for SNIM to make a substantial cor-;ribution (about US$34 million) towards the cost of the M'haoudat project. 138. SNIM has been making profits in 1989 and 1990 after making losses in 1987-1988 (see Annex) as a result of better iron ore prices and higher production levels. It has been able to bear the burden of heavy financial charges (interest and depreciation) i.e., US$86 million in 1989 and US$79.6 million in 1990. Prospects are that SNIM should remain profitable in 1991-92 but again incur losses in 1996-2000 because of the combination of still high (although declining) depreciation charges for Guelbs with new depreciation charges for the Mhaoudat project from 1993/94 on. - 31 - 139. Although SNIM is expected to have a negative net income during the second half of the 1990s, the long term financial outlook should improve since net losses would decline from US$15.3 million in 1995 to US$5.4 million in year 2000 is 3.5% of SNIM's total revenues because of a progressive decline of interest payments and depreciation charges (including M'haoudat) (cf. President's Report on a Development Credit to Mauritania for a Public Enterprise Sector Adjustment Program, May 30, 1990, paras. 39-41). Financial projections have been prepared in constant dollar terms because it tends to present a more conservative picture. It tends to somewhat overstate the weight of SNIM's debt service obligations. Consequently, one may expect that, with the coming into production of the M'haoudat Mine and assuming no major drop in iron ore prices, SNIM's financial performance would be sufficiently satisfactory to sustain a viable mining operation. 140. The likely sustainability of the SNIM Company, if not of the Guelbs Project as such, is important since SNIM remains a valuable assat for Mauritania, because of its capacity to generate foreign exchange for the Mauritanian economy, its ability to transfer technology to the whole industry sector, and to generate employment in the modern sector. 141. Even under the present tight financial situation, SNIM still generates about US$70 million foreign exchange after debt service and indirectly contributes about US$7 million to the Government's budget in the form of workers' income tax (PCR, para. 6.08). 142. Although the Guelbs Project performance was unsatisfactory overall, the variety of techniques involved in this project has been an enriching experience for Mauritania's mining and industrial sectors. As stated in the project completion review from the Borrower's perspective, "SNIM has acquired still greater in depth familiarity with: * different ore crushing and bonification techniques; * state of the art electronics and robotics; * mechanical assemblies from several different countries, based on a variety of technologies; * problems of starting-up complex industrial plants, etc." 143. While SNIM's total number of personnel is expected to stabilize at around its current level of 4,500 persons, it will continue to provide the living base for major portion of the population of Nouadhibou and Zouerate. About 200,000 persons, representing more than half the region's population, depend on it, directly or indirectly (PCR, para. 6.07). - 32 - VI. FINDINGS AND LESSONS OF EXPERIENCE Fdidngs 144. The cost of the Guelbs Project and the subsequent Rehabilitation Programs (1985 and 1987) has been high, i.e, more than US$620 million and the combined rate of return has been negative. Furthermore, SNIM has been burdened by high debt service and Mauritania, which has a very low per capita income, had to borrow heavily to pay for this debt. But the Guelbs Project was also financed at a time (1979) when Arab donors were seeking to assist Mauritania through large investment projects in order to channel massive funds lent on favorable terms. At the same time, European lenders (such as Caisse Centrale) and multilateral banks were also looking for project financing in order to sustain their lending programs and help in Mauritania's economic development. Few large project opportunities did exist but the Guelbs Iron Ore Project, although of dubious financial profitability, had the advantage for previous lenders to be a continuation of the earlier MIFERMA project, which they had successfully financed in 1960. The Bank and Caisse Centrale were consequently looking favorably at the prospect of financing an extension of the MIFERMA project. 145. The combination of an Arab desire to quickly channel funds to Mauritania and European/multilateral banks sympathetic attitude towards continuing lending :o a well known entity (SNIM had succeeded MIFERMA) led to the decision to finance the Guelbs Project. The Bank was put in charge of project evaluation and leadership in cooperation with the Kuwait Fund. The Bank, in addition to EIB and CCCE, wanted to pursue the "enclave" project approach which had been used for MIFERMA. The result was that the project was financed not with an IDA credit but through a Bank loan because large mining projects were at that time still assumed to be financially viable on those terms. Although understandable in the climate prevailing in the late 1970s and even perhaps in 1985 when the Rehabilitation Project was approved, the financing on relatively hard terms of a marginally profitable project proved disastrous and SNIM is still paying for such decision, despite further debt relief provided by the 1990 IDA Credit to finance PESAP. 146. The high risks due to exchange rate fluctuations were widely underestimated at appraisal. Such fluctuations have amplified the impact on SNIM's financial situation. For example, in constant terms, expressed in major European curreruies, the 1989 iron price was 40% lower than that of mid 1985. This has had a significant impact on SNIM's finances, since about two thirds of SNIM's supplies came from Europe (representing about 40% of its total cash expenditures), while its revenues are in U.S. dollars. 147. An additional problem for SNIM is that the debt is denominated in U.S. dollars or currencies which closely follow the value of the dollar. The other half is in currencies which, from 1987 to 1990, have generally appreciated vis-a- vis the dollar. 148. The lenders insistence on Security and Trust arrangements, although fully guaranteeing interest and repayment obligations, have forced SNIM to resort to overdraft financing and costly suppliers credits, thus building excessive - 33 - short-term debt. Trust at ..gements have also proved extremely complicated and hard to understand and foll-w (including by the Bank itself). 149. Another finding has been the somewhat excessive confidence put by Bank staff (who were responsible for the project appraisal and enjoyed the support of all other lenders) in the Engineers selected by SNIM to prepare and implement the Project. The risks of failure could have been minimized, if not avoided, if the detailed engineering design services had been performed by a firm which had more relevant experience and if a more appropriate mechanism than the so-called Technical Committee for review and control had been put in place. This Committee, although composed of internationally known experts, who, at the time, were the only ones to posses a practical experience of dry grinding process, was unfortunately unable to play its supervisory role fully (PCR, Part II, para. 15) .i' 150. After the poor start-up experience, it is now evident that more consideration should have been given to alternative process and plant designs with a view to eliminate or reduce the risks associated with the untried process and quasi-prototype plant. Such risks became particularly evident with the failure of the oxidized ore production line, the massive generation of dust throughout the plant and the mechanical problems experienced with the big grinding mills. 151. Because of the prototype nature of the bonification plant, extensive pilot plant testing was carried out. Nevertheless, the results of the commercial plant appear to indicate that the assumed scaling-up factor and the hardness of the ore were underestimated. In retrospect, a much more conservative approach in using pilot plant data would have been appropriate (PCR, para. 4.03). 152. An even more important and fatal mistake was to believe that no new rich ore deposits were available in the area.- No serious exploration work took place in the period 1975-1979 before the project was financed. And it is only in 1986, when the Project had put SNIM's on the brink of financial disaster, that SNIM found additional rich ore resources in the Kedia mines and at M'haoudat si EIB experts believe that the project management setup (SOCOMINE staff assisted by specialized engineering firms and controlled by a Technical Committee) provided good quality work and at a lower cost than an enginecring company who would have been selected on a cost-plus-fee basis. V EIB experts are of the opinion that, apart from problems due to scaling up factors from pilot plant testing and treatment of oxidized ores, conventional grinding methods would have been less efficient (overgrinding, costs). They agree, however, that problems would have been minized if three grinders of a smaller size had been installed. t There were few indications that rich ore was available, and most ore reserves consisted of low Fe content quartzite. A confusion was, however, made between the outside aspect of the rich M'haoudat ores (as identified later) and the aspect of magnetic rocks also located in the area. -34- simply in the latter's case by following up on minor surface indications of high grade ore which were already known in 1979. 153. Operating costs have been reduced in recent years but, perhaps due to weak project management, costs savings measures could and should have been sought long before the 1985 Rehabilitation Program. 154. Finally, the Bank's iron ore price forecL - ts used for appraisal of both the Guelbs Iron Ore Project and the SNIM Rehabilitation Project turned out to be far too optimistic, although they were considered conservative at the time. A more detailed analysis of future world supply/demand prospects for iron ore and steel based on closer contacts with other European and Japanese lenders might have led to more realistic forecasts. The Bank had been selected by all lenders as the chief operator for the Guelbs Project appraisal and this may have resulted in overconfidence in the Bank's ability to forecast commodity prices. Lessons of Experience 155. The following lessons can be derived from the Guelbs and SNIM Rehabilitation Projects: (a) Large mining expansion projects have to be undertaken only after a careful iron exploration program has been designed and executed. No preconceived assumptions such as the fact that previous exploration work has been undertaken in earlier years and no rich ore was found, should be taken as an excuse for not conducting further exploration work. Mining projects are usually highly risky and continuous, careful exploration program is a way to minimize such risks. The Guelbs project is another example of the danger for geologists to adopt excessively simplified conclusions. (b) The spme applies to such an important resource as water, particularly in a desertic environment. Systematic search for additional water resources, which were found later, would have totally changed technology choices for the Guelbs Project. (c) Reliance on pilot testing and expert technical committees is no safeguard against the risks connected with prototype plants. The decision to go ahead with such plants should not be made without full consideration of all elements involved: ore hardness; scaling up factors from pilot plant testing; availability of suitable, reliable equipment; Pnd experience with other prototypes elsewhere. In princip. i, the use of new untested technology should be avoided, pe.,;icularly in a difficult, desertic environment such as the oahara. Even today, experts remain divided on the question as to whether the magnetic separation dry process wa.s the best available. (d) When selecting equipment, maximum weight should be given to sturdiness even if capital costs seem relatively high, whenever this is compatible with the projects' finances. The Guelbs - 35 - Project was completed at a cost lower than originally estimated partly because too much stress may have been put on locating cheaper suppliers but the real cost resulting from equipment deficiencies has been extremely high. (e) Experienced operators from a similar industrial plant should be sought to design and implement a project. Before relying on engineers with integrity and professional competence in iron ore mining but not in complex bonification technical process, alternative solutions such as hiring a specialized engineering firm should be carefully evaluated. It is to be regretted that only after several years of bad experiences, SNIM hired a manager/engineer, who had a key role in a similar Canadian project. (f) When a project (such as Guelbs) is known at appraisal as having a very marginal rate of return, extra-conservative price assumptions should be made. Sensitivity analysis should show results obtained by using price forecasts below the official Bank's estimates and the final decision to finance the project should be made in giving utmost consideration to profitability obtained by using lower price projections. (g) Consideration should be given to the high risks resulting from foreign exchange fluctuations. Such risks may prove very high indeed depending on the denomination of the debt in various currencies and on the origin of equipment purchased. Allowance should be made for such fluctuations in sensitivity analysis and project risks calculations. (h) Special arrangements such as Security and Trust agreements should be simply designed and clearly understood by all parties concerned. (i) A coherent link has to be astablished between terms of lending on one hand and the financial prospects of the Project on the other hand. Lenders' understandable desire to be repaid fully and in time can only be frustrated if the burden of debt becomes unbearable for the borrower. The Bank and some other Europer lenders preoccupation with an "enclave" project approach assiste. by loans on relatively hard terms when the project was known to be very marginally financially viable, led to high debt service and finally to massive debt relief. The 1990 IDA credit was granted with the knowledge it would be largely used to help SNIM repay its earlier loans. Financing must be adapted to the characteristics of the project. (j) Financing large mining projects with a doubtful financial and economic justification should be avoided. An overly pessimistic view of the country's economic prospects may lead to major mistakes. The decision to go ahead with the Guelbs Project was made just before the fisheries sector started to develop to the - 36 - point of becoming more important in a few years than the mining sector itself. Fish exports have provided substantial foreign exchange receipts to Mauritania. Financing the Guelbs Project did not prove justified from a balance of payments point of view. - 37 ATTACHMENT I se agreement with PPAR sent to Borrower) FAX Ne 1 202 67 60 560 DEBTII EI I: MR. MARK BAIRD, CHEF DE DIVISION POLITIQUES ECONOMIQUES,INDUSTRIE ET FINANCL BENR2IRE: MR. MOHAMED LEMINE OULD DEIDAH, CONSEILLER ECONOMIQUE DU MINISTRE DU PLAN DE MAURITANIE P2: 7 AVRIL 1992 03I: PROJET DE MINERAI DE FER DES GUELBS ET PROJET DE REHABILITATION DE LA SNIM (PRETS 1747 ET 2643-MAU) MONSIEUR LE CHEF DE DIVISION, J'AI L'HONNEUR DE VOUS INFORMER QUE JE N'AI PAS D'OBSERVATIONS SUR LE RAPPORT D'EVALUATION RETROSPECTIVE DU PROJET EN OBJET, QUE VOUS AVEZ BIEN VOULU ME SOUMETTRE PAR VOTRE LETTRE EN DATE DU 18 FEVRIER 3992. LE RAPPORT PRESENTE EN EFFET A MON AVIS, LES PRINCIPAUX PROBLEMES AUXQUELS LE PROJET A FAIT FACE,ET TIRE LES CONCLUSIONS PERTINENTES DE CETTE EXPERIENCE. JE VOUS PRIE D'AGREER, MONSIEUR LE CHEF DE DIVISION,L'EXPRESSION DE MA HAUTE CONSIDERATION. NOXMD LI)IN OULD DEIDAN - 38 - ATTACHMENT II (These commenta are taken into account in the PPAR) 3 AR '92 BEEII LUXEMBOLRG FA: 4 El D Europake ivtoringbank % Europâie Investitionsnk EupWOaïNá Tpéna(a EnsvéûOiwv e European Investment Bank Sno Europeo de Inveruiones S'onque eurepéenne d'investinsenment Bance europea par gli inveslimenti Eurepes Invstering"sank Banque Mondiale Sanco Europou de investimento 1818 4th street N.W. Washington DC 20433 USA A l'attention de M. Mark Baird Luxembourg, la 10 avril 1992 PA/ACP2-C/EF.ms/4460 Oblet : Mauritanie - Projet de Minerai de Par des Guelbs et projet de réhabilitation de la SNIM (Prêt 1747 MAU et Prêt 2643 MAU). Rapport d'évaluation rétrospective du projet. Nous avons bien reçu votre lettre du 18 février et vous remercions de nous avoir transmis la version préliminaire du rapport d'évaluation du projet cité en objet. Il s'agit d'un rapport très intéressant si on considère la complexité que peut revêtir ce dossier et nous partageons dans l'ensemble bon nombre de ses conclusions et recommandations. Il nous parait toutefois souhaitable de vous faire part de quelques remarques sur des "interprétations" des auteurs du rapport qui nous paraissent incorrectes, ainsi que de notre opinion sur une pondération autre que celle du rapport de certaines causes des problèmes identifiés par rapport à d'autres. Elles se limiteront ici aux leçons à tirer de l'expérience (point 30 du sommaire, pages xiv et xv). Point 30 a) Nous pensons que l'application au cas considéré du principe énoncé, doit être nuancé. Les raisons pour lesquelles les recherches antérieures ont manqué de reconnaitre l'importance du Sisement de M'Haoudat sont explicables et paraissent compréhensibles, sinon rétrospectivement excusables g • abondance d'indices de Fe dans quasi toutes les collines situées dans un rayon de 80 km autour de Zouerate, tous indices ayant été reconnus j - constatation que parmi les indices reconnus seuls de rares cas isolés contenaient du minerai de Fe riche, la quasi totalité étant constituée de quartaite rubanni plus pauvre ; - confusion entre l'aspect lustré du faciès riche de M'HEoudat (tel qu'identifié plus tard) et celui caractéristique des roches magnisiennes également présentes dans la région; - l'éloignement du Guelb M'Haoudat de Zouet*te (60 km). ./. -39 - ATTACHMENT II Page 2 of 6 La conclusion que la région n'avait pas fait l'objet d'une prospection suffisante (Evaluation §112-113, §152) nous parait hâtive et exagérée. ou égard au très important effort de prospection réalisé. La principale leçon est à tirer par les géologues à qui ce cas donne un nouvel exemple du risque, permanent dans ce métier, de se laisser enfermer dans des schémas métallogénétiques exagérément simplifiés. Les Guelbs eux-momes ont fait l'objet d'une reconnaissance intensive ayant totalisé dans une première phase 40.000 a de sondages, ce qui était adéquat. Une deuxième phase a compris environ le double en sondages de préparation de l'exploitation. Point 30.b) Un traitement à l'eau, même si il s'était avéré réalisable. aurait sans doute été inapplicable un raison de l'opposition qui n'aurait pas manqué de se manifester de la part des milieux écologiques. Point 30.c) Cette opinion nous parait devoir être nuancée en rappelant que la choix du procédé s'est effectué sur base de plusieurs considérations s • les progrès réalisés en séparation magnétique permettaient d'envisager l'enrichissement du minerai des Guelbs par sép4ration magnétique. Les oxydes de fer présentaient en effet un caractère magnétique plus ou moins prononcé. C'était à l'instruction un des pointa les plus critiques ; • 1.'éloignement de toute source d'eau renouvelable imposait un traitement à sec (commentaire ci-dessus) s • la nature très dure du minerai permettait d'envisager un broyage autogène à grand diamètre, qui présentait également des avantages de coûts t • la possibilité de produire du sinter feed en une étape. Cette voie a été explorée de façon approfondie par des tests de broyabiliti et d'enrichissement dans un nombre important de centres de recherche minéralurgiques, suivis par des essais dans une usine pilote de 8 t/h construite à Zouerate. Environ 30.000 t de minerai de caractéristiques diverses ont été testées dans cette usine. Un soin particulier avait été apporté à l'étude de coefficients d'extrapolation des dimensions et des paramàtres de marche. Malgré qu'à l'époque les précautions prises nous avaient paru assurer la faisabilité technique des procédés avec un risque acceptable, il faut reconnaître avec le recul que 1'approche n'a pas dté totalement couronnée de succès. En particulier on peut admettre que l'évaluation, à partir des résultats de l'usine pilote, des coefficients d'extrapolation et des quantités de poussières générées n'a pas été suffisamment précise (Evaluation §151), m.is il n'est pas certain qu'elle aurait pu étre davantage approfondie avec les moyens d'investigation de l'époque. - 40 - ATTACHMENT II Page 3 of 6 13 APR '92 42& BEI'EIB LUXEMBOURG FAX4:437704 Lorsqu'on fait le compte des essais effectués, et notamment des plus de 4.000 h de fonctionnement de l'usine pilote, il nous paraît exagéré de dire que "la technologie utilisée n'a pas été entièrement testée" (Résumé §8). Il nous a paru et nous continuons de penser qu'il aurait été difficile de faire mieux en la matière. On retire du rapport (Evaluation §22-31) l'impression d'une critique que l'analyse de l'évaluation s'est arrêtée à une étape intermédiaire du processus de détermination de la technologie. Nous estimons que le choix du procédé, critiqué par le rapport (Evaluation §150), était et est probablement encore le meilleur disponible, mises à part les questions de taille et celles relatives au traitement des fines hématitiques : - le broyage conventionnel aurait été moins performant (surbroyage, coûts) ; - les broyages et séparations à l'eau, tels que suggérés par les auteurs du rapport, étaient radicalement à exclure, en raison de la rareté de l'eau, ou de sa nature saumàtre. On aurait cependant vraisemblablement fait l'économie de problèmes (mais non de coûts) en installant 3 lignes de broyeurs autogènes de moindre dimension. Point 30.d) Nous sommes d'accord. Point 30.e) (ainsi que Résumé §8 ; Evaluation §21-22, §42 et §149) Nous ne pensons pas que la conception de la gestion du projet ait été mauvaise. La structure de gestion du projet, constituée de l'équipe Socomine, renforcée des ingéniéries spécialisées, et contrôlée par un Comité Technique, a fait un travail de qualité, dans un délai plus court, et certainement à un coût moins élevé que si l'autre solution proposée (société d'ingéniérie en cost plus fee) avait été retenue. Cette structure se justifiait aux yeux des équipes d'instruction par la connaissance que Socomine avait de la société, par une familiarisation plus rapide du personnel de la SNIM avant le démarrage, par des problèmes qui n'auraient pas manqué de se poser entre la société d'ingéniérie et la SNM, ainsi que par les délais et coûts qui s'en seraient suivis (1). Un contrat clé-en-main ou un contrat d'ingénierie unique avec des sociétés du type Bechtel, Davy McKee ou Sofresid n'aurait certainement pas pu permettre d'éviter les problèmes qui se sont posés au niveau du process. ./. (1) La solution société d'ingéniérf adoptée dans un autre projet minier en Mauritanie (SOMIMA) financé par la BEI avait donné de très mauvais résultats. - 41 - ATTACHMENT II Page 4 of 6 13 APR '92 te: BEI/EIB LUXEMBOURG FAX4:437704 Si Socomine n'avait effectivement pas l'expérience minéralurgique requise, elle s'est cependant associée avec Sofresid et Fives-Cail-Babcock, des sociétés expérimentées dans ce secteur. Il n'est pas exact d'affirmer, comme le fait le rapport (Evaluation §42) que FCB n'avait aucune expérience du traitement minerai de fer et de ce procédé. FCB a réalisé plusieurs usines de traitement et avait en outre l'expérience de broyeurs autogènes de dimension similaire pour l'industrie cimentière. Ce choix avait été approuvé par le Comité Technique, dont un membre avait vécu l'expérience de Carol Lake (avec un autre fournisseur). Le Comité Technique apportait l'expertise des seules perc,,es- qui à l'époque avaient une connaissance concrète du broyage autogène à sec. Il est injustifié et même provocateur de dire que ces experts servaient de couverture au bureau d'études (Evaluation §149), étant donné la nombre de recommandations utiles qu'ils ont émises. Par contre on peut souligner des lacunes au niveau de la structure de démarrage de l'exploitation, où une direction unique aurait été nécessaire à plein temps, renforcée de personnes expérixantées dans ce type d'exploitation (qui ne sont arrivées que plus tard). Les flottements constatés pendant le démarrage de l'usine et durant les premiers temps de l'exploitation auraient probablement pu être atténués. Il a manqué d'une poigne ferme pour diriger l'ensemble et éviter la confusion résultant des demandes conflictuelles de l'Exploitation, qui voulait aguerrir son personnel, de l'Entretien, qui voulait assurer le bon état de l'outil, du Process, qui voulait déterminer les paramètres de marche pour les différents types de minerais, et de la Commercialisation, qui réclamait des produits à vendre, - tout cela pendant une période où les réceptions provisoires de certains équipements clés n'avaient pas encore été effectuées. Il a fallu trop de temps à la SNIM pour se rendre compte de la gravité de la situation et de l'impasse dans laquelle elle s'acculait et sans doute également aux bailleurs de fonds pour en apprécier les conséquences. Point 30.f) Nous sommes bien d'accord, l'opinion émise ici rejoint nos propres sentiments. Point 30.a) et h) Nous sommes d'accord dans une certaine mesure, ces points ayant cependant été pris en considération, mais de facon peut être inappropriée, lors de l'instruction du projet. Point 30.i) ainsi que (Résumé §30i, Evaluation §155) Nous sommes en désaccord complet avec l'opinion que la volonté première des bailleurs de fonds était "de faire un gros projet, et puisque SNIM Guelbs était le seul gros projet à l'horizon on a fait celui-là". La réalité de l'époque par contre était que : • la SNIM ézait de très loin la principal agent économique du pays, l'alternative constituée par la pêche étant - 42 - ATTACHMENT Il Page 5 of 6 13 APR '93 14:30 BI.EIB LUXEMBOURG FAX4:437704 considérée, comme très difficile à mettre en oeuvre, sujette à irrégularités peu contrôlables, et de toute façon une entreprise de très lQngue haleine ; il existait d'amples réserves de minerai aux alentours de Zouerate, pour lesquelles on estimait avoir trouvé un procédé de traitement viable ; sans la réalisation du projet des Guelbs, il existait un risque sérieux pour la SNIM de voir décroître et arrêter son activité minière (l'ancienne Somima - exploitation de cuivre - avait déjà été arrêtée vers 1979), privant le pays de ses principales ressources en devises, et d'un pôle de développement importantt - des raisons géopolitiques recommandaient sans doute d'éviter un risque de suppression d'emplois et de propagation d'une instabilité dans cette région. Si ces deux derniers points ont peut-être pu influencer les instances de décision de certains des bailleurs de fonds, ils n'ont pas affecté le jugement tachnico-économique des équipes d'instruction du projet. Nous n'avons pas eu non plus le sentiment d'une hâte excessive dans la décision de financement, mais bien au contraire de lenteurs, qui ont fait craindre à plusieurs reprises que la soudure entre exploitation de la Kédia et des Guelba ne puisse se faire de façon harmonieuse. Ultérieurement il est apparu, part suite d'éléments non prévisibles lors de la mise en place du projet, que ces craintes étaient non fondées. Enfin, les ptints suivants ont également attiré notre attention Exécution du projet Tel que nous les avons estimés, les retards mentionnés dans le rapport (Résumé §16 ainsi que Evaluation §82 et suiv.) auraient plutôt eu les causes suivantes : Retards dans l'achèvement • mise en place du financement (accords trust-deed, cross-effectiveness) ; - certaines longues négociations de la Direction de l'entreprise avec les fournisseurs (notamment le mauvais contrat Krupp) ; - retards et erreurs graves des fournisseurs (notamment Krupp). Retards dans la montée en cadence : - modifications d'installations déficientes par les fournisseurs ; - organisation multicéphale lors du démarrage - lacunes dans la c-ordination de quelques intervenants ; - problèmes de process. En particulier, les retards dus à la structure de financement nous paraissent sous-estimés. - 43 - ATTACHMENT II Page 6 of 6 13 APR '92 J,3J BEI'EIB LUXEMBOURG FAX4:437704 Suivi du Projet (Evaluation §153) L'énumération en tête du rapport des jours de mission de la Banque Mondiale suffirait à démentir qu'il y a eu au moins de la part de cette institution un manque de suivi. Nous pouvons indiquer qu'il en a été de même pour les autres bailleurs de fonds et notamment la BEI. Ceux-ci ont réagi et réclamé des mesures de redressements et d'austérité dès 1982/83. Il faut préciser que la tâche des bailleurs de fonds a été ensuite facilitée par la prise de conscience et la détermination de la nouvelle direction de la SNIM et du gouvernement mais à partir de 1986. Perspectives (Résumé §25; Evaluation §135-136) Nour partageons l'opinion que le plan de réhabilitation a été globalement un succès, bien que là aussi des retards se soient produits et que les effets du plan ne se soient pas encore tous traduits en termes de production. Nous sommes également d'accord pour considérer que le nouveau gisement de M'Haoudat permet à l'entreprise d'envisager l'avenir avec plus de sérénité. Le point essentiel qui nous parait ressortir de cette expérience concerne la nécessité d'une adaptation du financement aux caractéristiques du projet, ce qui semble avoir été mieux pris en considération pour le projet M'Haoudat. Veuillez agréer, Monsieur, l'expression de nos sentiments distingués. BANQUE EUROPEENNE D'INVESTISSEMENT J. Silvain J-J., Mertens ANNEX MAURITANIA - SNIM- HISTORICAL INCOME STATEMENT (1971-1990) (US$HILLION) 1971 1975 1976 1977 1978 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 I. Sales Volume (Million tons) 8.60 9.65 9.66 8.42 6.47 8.85 7.63 7.38 9.50 9.33 8.93 9.00 10.00 11.14 11.35 II. Revenues Iron Ore Sales 4.35 6.77 7.03 5.91 4.26 8.88 9.61 7.76 10.37 150.70 142.40 135.90 137.00 172.10 207.50 Others 0.36 0.84 0.54 0.46 0.35 0.48 0.22 0.22 (0.13) 9.80 3.10 11.90 14.90 36.10 49.80 Total Revenues 4.81 7.61 7.57 6.37 4.61 9.36 9.83 7.98 10.24 160.50 145.50 147.90 151.90 208.20 257.30 III. Costs Cost of Sales 2.70 4.92 4.85 4.86 3.70 6.70 7.20 6.27 7.75 117.70 107.00 107.10 110.00 117.70 129.50 Depreciation 1.34 1.03 0.96 0.93 0.94 1.01 1.07 1.28 1.31 21.50 15.60 21.00 15.80 45.80 34.20 Operating Profit 0.77 1.66 1.76 0.58 (0.03) 1.65 1.26 0.43 1.18 21.30 22.90 19.80 26.10 44.70 93.60 IV. Financial Charges 0.22 0.22 0.21 0.19 0.22 0.16 0.14 0.15 0.15 1.80 10.70 25.70 25.40 41.00 45.40 V. Profit Before Taxes 0.55 1.44 1.55 0.39 (0.25) 1.49 1.12 0.28 1.03 19.50 12.20 (5.90) 0.70 3.70 26.70 VI. Taxes 0.39 0.45 0.71 0.59 0.19 0.46 0.43 0.40 0.93 13.70 11.70 5.50 5.50 - - VII. Profit After Taxes 0.16 0.99 0.84 (0.20) (0.44) 1.03 0.64 (0.12) 0.10 5.80 0.50 (11.40) (4.90) 3.70 26.70 VIII. Net Profit/Iron Sales (1) 3.60 14.60 12.00 (3.40)(10.30)11.60 7.20 (1.60) 1.00 3.80 0.40 (8.40) (3.60) 2.15 12.90 UT Source: SNIM - Financial Statements