simio competition october 2016

27
1. Problem Statement 2. Methodology 3. Scenario detail Local DC with Periodic review Regional DC with Periodic review Local DC with Continuous review Regional DC with Periodic review 4. Assumptions 5. Simulation Logic 6. Scenario selection 7. Result tabulation 8. Result Analysis 9. Conclusions Outline

Upload: abhishek-vaid

Post on 15-Jan-2017

148 views

Category:

Documents


13 download

TRANSCRIPT

PowerPoint Presentation

Problem StatementMethodologyScenario detail Local DC with Periodic review Regional DC with Periodic review Local DC with Continuous review Regional DC with Periodic reviewAssumptions Simulation Logic Scenario selectionResult tabulationResult Analysis Conclusions Outline

1

Problem StatementAbout Company : Company has operations in Central America, South America and Caribbean. Central America Costa Rica , Guatemala , El SalvadorSouth America Uruguay , ParaguayCaribbean Jamaica

Key DeliverablesDetermine which Inventory policy should be selected for Scenario 1 &2 :Scenario 1 is whether to have a local distribution center in retailing countries Scenario 2 is whether to have a regional distribution center at Miami or Panama for expansion strategyAnalyze the service level for both Scenario 1 and Scenario 2

Inputs Available :Forecasted Demand, various required costs, lead time , Unloading time, ships departure and capacity Key Performance Indicators to evaluate Profit Service level Safety Stock levelManufacturing/Logistic cost Sales Revenue

Methodology

This is a warehouse distribution problem and the theoretical concepts needed stem out of inventory control.There are two inventory models discussed in the problem statement:Periodic review-The inventory on hand is counted at specific time intervals; for example, every week or at the end of each month. If the inventory in stock is determined to below a certain level ( reorder point) , an order is placed for an amount that will bring inventory back up to a desired level.Continuous review -A continuous record of the inventory level for every item is maintained. A positive feature of a continuous system is that the inventory level is continuously monitored, so management always knows the inventory status. As soon as the level falls below reorder point , an order is placed for order up to level or of an economic order quantity (EOQ).EOQ is one of the two main values needed in inventory control, the other being reorder point. Reorder points are calculated using product lead time and safety stock desired to achieve desired service level. Service levelis the probability of not having stock outs. It is a KPI that is verified in simulation results. It is calculated using the formula Where ,A= Setup Cost/per orderiC = Holding cost/unit timeS= Demand/Unit time

Example of Inventory Models

Below shown models are a good depiction of how inventory control can be done for this problem.

Fig 1. shows that an order is placed as soon as inventory reaches r and is how our continuous model also operates.

In Fig 2. an order is placed to order upto level at start of each review period. Our periodic review inventory graph is similar to this.

Figure 1 Figure 2

Methodology Cont...

Going into operation detail , the manufacturing base is in Asia and end product is shipped to respective companies. The order can be shipped on a weekly basis if necessary and there is no wait time as the inventory at Asia is more than demand .

The model needs to incorporate country specific ordering pattern and cost pattern. Analysis can be done on regional basis subsequently.

The reorder point and order quantity for each country needs to be determined. These values will also be subject to optimization using SIMIOs Opt quest capabilities.

Transport cost is a major factor that is not incorporated into EOQ calculation and is used during simulation runs to maximize profit.

As far as operations are concerned , the data is not available s supposed to be assumed in a realistic manner and these assumptions are stated clearly is subsequent slides.

ScenariosAs mentioned before , there are four initial scenarios that were created and analyzed.Data is available for these in form of transport time , transport Cost form Asia to specific region , daily demand/country , operating cost , manufacturing cost , sale cost and inventory delays/damages . 1 & 2 .) Local Distribution Channel with Periodic review /Continuous review

The Supply for this has Local DC in each country and flow of material is directly from Asia to respective country.Separate Inventory model have to be created w.r.t Review period , order quantity and reorder point.For the continuous model , review period is not an inventory control and is replaced by an event based decider which is triggered by inventory status

Central AmericaManufacturing Setup in AsiaSouth AmericaCaribbean

Central AmericaManufacturing Setup in AsiaSouth AmericaCaribbean3 & 4.) Regional Distribution Channel with Periodic review/Continuous reviewRegional DCThe Regional DC which can either be in Miami/Panama is the supplier for all three regions , it has a separate supply chain with Asia and thus has its own inventory model that needs to be defined. The transport cost and transport time are consequently changed depending on route selection.

Major difference between this scenario and earlier scenario is the addition of inventory variation at Regional DC i.e. unique reorder point , order quantity and review period.

As only one option , Miami or Panama can be selected , the model had the capability of selecting one of them prior to the run.

The continuous model employs the same logic as the earlier discussed one and is more complex due to the addition of an extra object in the form of a regional DC.

AssumptionsThe critical assumptions that are taken for this problem are :

No delay in processing daily demand orders i.e. fulfilled the same day.

Asia manufacturing plant has sufficient inventory to not have any backorder.

Infinite number of ships available at Asia dock , we have restricted the number to 2 per route as it is a sufficient number to not have order waiting at a dock.

Order processing and loading time at a dock is one day.

Order processing time at Asia is half a day.

An order is sent to lost sale if it cannot be met fully.

Regional DC have initial stock level as a sum of initial inventory at each country.

Backorders are not considered.

TEU are rounded up for calculating transport cost.

Assumptions not listed here would be minor in terms of functionality and would be mentioned at appropriate place in the model .

Simulation Logic- Demand Fulfillment Demand arrives daily and order get processed if initial inventory is higher than arrived country demand

In a periodic review , Inventory level is checked every review period and order up to quantity can be only order at review period only.

In a continuous review , when Inventory on hand reaches to reorder point then fixed and optimal order quantity is processed to meet country uncertain demand.Sufficient Inventory on hand result demand met for market and sometimes long lead times results stock out conditions and company lose the sales which can be generate to revenue

Daily DemandInventory level > Demand Update Inventory , Update Sale CostYes Demand metFalse Sales LostUpdate lost sale Cost , Update service levelDemand LogicOrder processing

Simulation logic Demand fulfillment

Simulation Logic- Replenishment Order Placement Processed at AsiaUpdate InventoryFor Local DCOrder Placement Processed at Regional DCUpdate Local DC InventoryRDC Inventory controlFrom Demand outputFrom Demand outputProcessed at ASIAUpdate Regional DC InventoryReplenishment Timer is triggered if Status Inventory is lower than Reorder point. Triggering event is review period in periodic review and for continuous , it is a real time check of inventory every day.For Regional DC

Simulation logic Replenishment

The above shown logic is how a replenishment order is placed at a regional DC. It has two branches for the two regional DC options and one of them can be selected from facility window prior to running the program.

Status of order This command is used to measure total inventory on hand so that over order is not done while inventory is in transit. It is used in the triggering event condition for this process

Simulation logic Cost LogicThree major costs which play pivot role in evaluating which Inventory model is beneficial for the company : Sales RevenueManufacturing Cost Ordering cost, Operating cost, Transportation CostHolding Cost or Inventory Carrying Cost

Define Number of shipments made TEU_CR= TEU_CR + ModelEntity.DemandCR / 1000Manufacturing cost is function of Ordering Cost, transportation Cost, operating cost at Asia and Local Retailers Holding Cost is a function of average inventory on hand and unit cost for holding it.

Model Verification - IThe model was developed in three sections demand fulfillment , replenishment and cost for manufacturing which includes, transportation and operating cost and holding cost to maintain the inventory on hand.Number of cases expected from each country are verified as below

Color coding, animation used to observe the Inventory on Hand update once order is arrived to the country. Ships are used as a vehicle in order to load and unload the product from Asia manufacturing to the market or Asia to Regional Distribution channel.

CountryAverage Demand/DayRun Length ( Days)Expected DemandObserved DemandError (%)Costa Rica7153362402402331072.97Guatemala11433363840483863460.60El Salvador10713363598563572860.71Uruguay7153362402402352382.08Paraguay5363361800961766161.93Jamaica7153362402402449031.94

Model Verification - IIThe model was developed in three sections demand fulfillment , replenishment and cost for manufacturing which includes, transportation and operating cost and holding cost to maintain the inventory on hand.Number of cases expected from each country are verified as below

Inventory Dashboard verifies that Inventory on Hand is updated once arrived from Asia manufacturing or Regional DC to Local Retailers Sales Dashboard helps to observe the number of sales have been made to meet customer demand and how many sales are lost due to unavailability of product in stockCost Dashboard provides overall view to cost aspects. This dashboard explains which cost is playing major role in the profit.

Scenario selection

Opt quest was used to run experiments which had KPI as responses . There were a lot of controls defined and specific controls were selected for each run.

Common controls are reorder point , safety stock level to both scenarios and specific ones like order upto quantity for periodic review

It was programmed to create upto 50 different scenarios and 95 % confidence interval for the results

OptQuest gave the capability of selecting the best scenario based on a multi objective response

As Profit was the primary response , results were tabulated according to it and other KPI analyzed subsequently

Model Video

Scenario 1 & 2 Result Discussion Opt quest provided best Service level by optimizing Initial Inventory , Reorder point , review period and Order up to quantityReviewContinuousScenario1Asia Direct To MarketHighly competitive Service levels in Central America and CaribbeanLoss of sale in South America due to long lead time Best Scenario among all 4 in terms of Demand fulfillmentScenario2Asia To Regional Distribution To MarketDemand not satisfied because of less inventory on hand and delay in replenishments Highly competitive Service levels in all regions Slight Loss of sale in Central America and Carribbean

Scenario 2 with review policy is rejected due low service levels. All others will be analyzed further.Inventory Policy Scenario

ReviewContinuousScenario1Asia Direct To MarketProfit Margins are the highest in this Case for all regionsOptimized review period indicates less loss of SalesProfit Margins are competitive w.r.t to Period Inventory modelPolicy can be implemented only for South AmericaScenario2Asia To Regional Distribution To MarketHigh loss due to increased operating cost and transportation cost due to addition of Regional DCLow sales observed compared to other Inventory Models

Inventory Policy Scenario

Details are added in Next SlidesScenario 1 & 2 Result Discussion

ReviewContinuousScenario1Asia Direct To Market6% more demand has been fulfilled resulting higher transportation cost and ordering costLow sales due to less stock on hand as no safety stock at Local RetailScenario2Asia To Regional Distribution To MarketAdditional operating costs and transportation cost resulting almost 50 % less sales than other models

Inventory Policy ScenarioScenario 2 with Continuous policy is rejected due to high manufacturing cost and high overheads .

Scenario 1 & 2 Result Discussion

ReviewContinuousScenario1Asia Direct To MarketHolding cost are low due to optimal order quantity and high service levels, optimal review period resulting on time replenishments arrivalHigher Inventory Carrying Cost resulting lesser profit margins Scenario2Asia To Regional Distribution To Market

Inventory Policy Scenario

Scenario 1 with Review policy is best Inventory model w.r.t to all key parameters for further evaluation Scenario 1 & 2 Result Discussion

Best Scenario Result and DiscussionBest Scenario is Scenario 1 which represents direct flow of products from Asia manufacturing to Local Retailers and it should follow the review periodic model.Results below are discussed in detail only for the best scenario :Initial Setup for Reorder point, Initial Inventory and Order upto quantity has been calculated by using EOQ , Base stock (r,R) modelPerform QptQuest with theoretical values of reorder point, initial inventory and order up to quantity as lower bound and upper bound +10,000 to lower boundSafety stock Level has been varied in order to perform an assessment on sales revenue

Country Demand/DayA M BMean demand per daySetup Cost Holding Cost per day EOQReview Periodsafety stockOptimal Solution with 0.15% safety stockReorderOrderUpToQInitial InventoryCosta Rica0715143071554500002019740283450227554249519550Guatemala011432286114354500002024959225520363816134031280El Salvador010712142107154500002024160235175340925825229325Uruguay0715143071554500002019740286900545677430739100Paraguay0536107253654500002017092325175409085800029325Jamaica0715143071554500002019740283450227554249519550

Optimal order quantity has been identified in the best scenario and company should order upto this quantity to reduce loss sales and maximize profit

CountryManufacturing cost ( *10^9)Mean Holding( *10^6)Total Cost per Year (10^9)Profit (*10^9)Costa Rica20.379.320.436.12Guatemala20.4119.120.569.16El Salvador14.975.91566.50Uruguay20.892.72129.81Paraguay17149.41717.44Jamaica14122.81440.44

Best Scenario Result and Discussion-Part aBest Scenario is Scenario 1 which represents direct flow of products from Asia manufacturing to Local Retailers and it should follow the review periodic model.From the optimal solution, the total average cost per year and profits are evaluated as below

Average Total Cost per year to run the distribution is CRC 108.30 *109 which is ~21,60,78,400 USDProfits are estimated to be around CRC 260 *109 which is ~ 51,89,08,370 USD

Safety Stock levelProfit ( *10^9)Lost Sale cost ( *10^9)15%259.4529.3020%254.6136.7025%253.1837.12

Best Scenario Result and Discussion-Part bProfit lost at 15% safety stock is CRC 29.30 *109

Best Scenario Result and Discussion-Part c &dCountryReview periodTotal cost ( *10^9)Mean Total cost( *10^9)Sale Revenue( *10^9)Lost sale revenue( *10^9)Holding Cost (*10^6)Mean HoldingProfit (*10^9)Costa Rica2720.3 +/- .420.356.493.4879.3 +/- 3.279.336.12Guatemala2720.4 +/- .420.489.683.44119.1 +/- 4.2119.169.16El Salvador2714.9 +/- .314.981.473.5975.9 +/- 2.975.966.50Uruguay2920.8 +/- .320.850.708.3892.7 +/- 1.792.729.81Paraguay2717 +/- .81734.597.88149.4 +/- 4.6149.417.44Jamaica2714 +/- .21454.572.53122.8 +/- 3.5 122.840.44

CountryReview periodTotal cost ( *10^9)Mean Total cost( *10^9)Sale Revenue( *10^9)Lost sale revenue( *10^9)Holding Cost (*10^6)Mean HoldingProfit(*10^9)Costa Rica2719.8 +/ -.419.854.326793.7279.5 +/- 3.279.534.45Guatemala2720.1 +/- .420.186.2196656.77119.3 +/- 6.3119.366.00El Salvador2714.6 +/ - .214.678.3051856.9676.3 +/- 3.476.363.63Uruguay2920.7 +/ -.320.749.785968.0692.5 +/- 2.792.528.99Paraguay2716.3 +/- .9516.335.3449257.76190.4 +/ - 12.5190.418.85Jamaica2713.9 +/ - .213.956.7035353.43122.1 +/- 3.8122.142.68

Safety Stock 15 %Safety Stock 20 %CountryReview periodTotal costMean Total cost( *10^9)Sale Revenue( *10^9)Lost sale revenue( *10^9)Holding Cost (*10^6)Mean HoldingProfit(*10^9)Costa Rica2720.3+ /- .420.355.874.6886.5 +/- 3.186.535.48Guatemala2720.7 +/ -.420.787.414.97118.5 +/- 4.4118.566.59El Salvador2714.8 +/ -.314.878.099.4677.4 +/- 3.177.463.21Uruguay2920.9 +/ -.420.948.518.3293.8 +/- 293.827.52Paraguay2717.7 +/- .717.737.437.08147.1 +/- 4.3147.119.59Jamaica2713.9 +/ - .113.954.822.60122.3 +/- 3.6122.340.80

Safety Stock 25 %Profit lost at 15% safety stock is CRC 29.30 *109

Best Scenario Result and Discussion-Part eDemand FluctuationDifferenceOriginalSale madeSale lostManufacturing Cost Sale revenueProfitSale madeSale lostManufacturing Cost Sale revenueProfitCosta Rica5505717987381400000013488965000967496500018.35308929025100000000130068050007906805000Guatemala10633798553298000000260525650002275456500021.696426779557800000002362437000017844370000Uruguay392660351400000096201700006106170000-63.15792932074233000000141926050009959605000Paraguay267700385900000065586500002699650000-58.34247206133000000104056400004272640000412353500003.0439983420000

Demand Fluctuation for 8 weeks Costa Rica and Guatemala up by 30 % Uruguay and Paraguay down by 50 %Run length 12 weeks

Profit lost at 15% safety stock is CRC 29.30 *109

Conclusion Company should select scenario 1 of having a local distribution channel with Periodic review policy.Expansion strategy at Panama or Miami as regional DC will not be a good strategy as it will only incur higher loss sale and less customer satisfaction.Company can define individual review period according to the forecasted demand for each country.Model indicates that transportation cost and operating cost are high due large sale volume and consequently increases manufacturing cost.Recommendation Company should definitely think about implementing periodic policy only in Central America and Caribbean and continuous review policy to South AmericaThere should be higher safety stock at Uruguay and Paraguay as their lead time to receive replenishment is quite longTransportation cost for South America should be negotiated or shipping vendor should be changedRegional DC might be helpful but other locations should be identified than Panama and Miami