Inventory Distribution Model

This project represents the investigation into the interactions associated with an endeavor to reengineer an inventory management and distribution operation.

Supporting Analysis

Architecture of the System

Part of the reason we often have such difficulty effecting the chances we set out to make has to do with our limited understanding of the systems we set out to change. We have a very pronounced tendency to zero in on one or two factors which we consider to be the root of the problem we face, or the controlling factors of the change we wish to effect. Yet, we so often fail to realize the extent of the interconnections and interactions within the system we attempt to change.

The following diagram represents the major interactions which are part of the supplier / distribution system associated with the LAA Group.

The standard response to diagrams of this type generally borders on something akin to horror, and the comment is that it's so complicated. It is our unwillingness to understand and work with the complexity of systems which dictates that we perceive them as complicated, and subsequently become the victim of them.

The components of the above diagram are described as follows:

Current Sales

Represents the current level of sales in units and $.

Current Sales is directly dependent on Demand. As Demand increases Current Sales will increase as long as there is Inventory available to fulfill the Demand.

Lead Time

The current delivery delay among the suppliers varies from 1 month to 5 months in some cases. The overall average delay is figured to be 4 months.

The belief is that by carrying the inventory in central location and accepting forecasts once a month, accepting orders every two weeks, and shipping every two weeks, via Air Freight, the average delivery delay could be cut to 2 weeks.

Demand

Demand is consider to drive Current Sales. As it is believed that Demand is depressed because of less than adequate availability then an increase in Perceived Availability will allow demand to rise back to the appropriate market level. Perceived availability will not influence Demand to actually grow, but rather just return to it's potential level based on market size.

This would tend to imply that as Perceived Availability increases Demand will approach some predefined limit and not grow beyond that limit unless some other factor enters into the puzzle.

Desired Sales

Represents the desired level of sales in units and $. This is the driving element in the whole equation.

Inventory

Inventory is the total level of inventory held in all the countries based. This is a result of increased based on shipments received and decreases based on Current sales.

Orders

Orders represent the total of orders placed against suppliers to replace current sales and satisfy future current inventory.

There are a couple questions regarding the interplay of Service Level and Delivery Delay and Orders. I am uncertain as to whether there is a direct influence between Orders and Service Level and Delivery Delay or whether Service Level and Delivery Delay operate on Orders via Required Inventory and Sales Forecast. I have to ponder this one some more.

Perceived Availability

Perceived Availability represents the degree to which the market believes the Inventory is available to meet their needs. To the extent there is Perceived Availability then the market produce Demand for the product. To the extent the market does not believe the Inventory is available it will attempt to fill its need via some alternative product.

Required Inventory

Required Inventory is the future inventory it is believed must be in place to respond to future demand at a point in time in the future equivalent to the Delivery Delay time.

Sales Forecast

Sales Forecast is the estimate of the sales expected in the future period equal to the delivery delay. This is essentially the amount of inventory that must be on hand at that time in the future to satisfy the Demand that will be experienced in that time period.

Service Level

Service Level is the percentage of the order received when the shipment arrives. Typically Service Level has been between 40% and 60%. For model calculations we have been using 50% as the typical average Service Level.

The Service Level is a function of the structure of the system and the operation of the suppliers to the system.

Shipments

Shipment represents the fulfillment of a prior order at the end of the Delivery Delay period.

The Shipment will consists of that quantity of ordered material based on the Service Level for the particular supplier.

Structural Change

The difference between Current Sales and Desired Sales, with Desired Sales being greater than Current Sales, provides the motivation for structural change based on the believe that structural change will influence the improvement in certain elements of the system which will result in future decreases in Inventory and increases in Current Sales.

It is believed that by inventorying in a central location and air freighting products to the individual locations, with the same level of demand, the total inventory in the system can be reduced. This based on the added belief that by carrying inventory in a centralized location Service Level and be increased and Delivery Delay can be reduced.

Governing Logic

Return on Inventory Assets = Gross Profit($) / Sales($)
Gross Profit = Sales($) - Cost of Goods Sold($) - Inventory Carrying Cost($)
Sales($) = Sales * (1 + GSV)
Order Quantity = Target Inventory - Availability
Target Inventory = Forecast * (Lead Time + Safety Stock + Order Freq)
Availability = On Hand + On Order - Backorder
Safety Stock = 2 * Order Freq * ( (1 - Fcst Acc) / Fcst Acc + (1 - Service Level) / Service Level )
Cost of Goods Sold = Product Cost * (1 + Landing %)
Inventory Turns = Average Inventory On Hand / Yearly Sales

Initial Year Collected Data - 1995

Sales - 5.84 million units
Inventory On Hand - 4.99 million units
Service Level - 60%
Lead Time - 6 months
Forecasting Accuracy - 25% (75% Inaccuracy)
Sales per month distribution
- Jan = 0.048
- Feb = 0.070
- Mar = 0.095
- Apr = 0.080
- May = 0.073
- Jun = 0.073
- Jul = 0.073
- Aug = 0.078
- Sep = 0.098
- Oct = 0.098
- Nov = 0.106
- Dec = 0.109
Product Cost = $.81
LAC Landing Costs = 32%

Parameter Values and Projections

Annual Growth Projections
- 1995 - 00.0% (actual base year)
- 1996 - 61.5%
- 1997 - 28.6%
- 1998 - 23.2%
- 1999 - 20.2%
- 2000 - 20.0%
Inventory Carrying Cost = 20% of Product Cost per Year
GSV = 43% of Product Cost
LAC Forecasting Accuracy = 25%
LAC Lead Time from Suppliers = 6 months
LAC Service Level from Suppliers = 50%
LAC Lead Time from MRDC = 1 month Ocean/.5 month Air
LAC Service Level from MRDC = 85%
MRDC Forecasting Accuracy = 70%
MRDC Lead Time from Suppliers = 4 months
MRDC Service Level from Suppliers = 85%

Simulation Runs

Initial Year Model Validation

For the model to be valid it must produce results which agree with the 1995 actual values. The following 12 month run is with no default parameter changes to produce correct 12 month sales volume, dollars, and end of year inventory.

Projection Runs

A 36 month run with no growth shows the expected inventory oscillation.

Note the stabilization of the yearly sales volume (line # 5) after the 12th month. The dip in return on inventory assets in month 13, 25, 37 are due to the lower sales volume which occurs between December and January, while inventory is increasing.

The following graph is produced with the projected growth factors turned on for the 2nd and 3rd years.

Note that inventory and yearly sales volume continue to grow as sales increase, yet the return on inventory assets remains at essential the same average level. This is because the ratio of gross profit to profit does not change with the growth in sales.

MRDC Operation

When MRDC operation is turned on there are a number of options that can be set at the same time that it is turned on. These have to do with the expected supplier lead time and service level to MRDC, as well as the MRDC lead time and service level to LACs. In addition the expected alteration in forecasting accuracy has a substantial effect on the inventory levels as does the method of shipment from MRDC to LACs.

The following graphs represents the implementation of MRDC operation in month 16, April 1996, with shipments to LACs by ocean.

Note that while annual sales is not affected there is a marked decline in LAC inventory during the 16th thru 24th months. Along with this there is a marked increase in return on inventory assets. The return on inventory assets increases because the decrease in inventory carrying cost more than offsets the increase in shipping costs from MRDC via ocean once a month.

If MRDC decides to ship via air every 2 weeks the implications are shown in the following graph:

Note that shipping by air every two weeks produces a lower level of inventory the return on inventory assets drops because the reduced inventory carrying costs are not enough to offset the increase shipping costs for air.

Some of the above relationships are best seen in comparative graphs, a number of which are provided below.

This graph represents resultant LAC inventory with (1) no growth, (2) growth, (3) MRDC operation with growth and shipment via ocean starting in month 16, (4) MRDC operation with growth and shipment via air starting in month 16.

This graph represents resultant system (LAC + MRDC) inventory with (1) no growth, (2) growth, (3) MRDC operation with growth and shipment via ocean starting in month 16, (4) MRDC operation with growth and shipment via air starting in month 16

Note the delay in the drop of the MRDC inventory until after the orders in the pipeline are received and redistributed to the LACs as they place orders.

This graph represents resultant LAC average inventory turns with (1) no growth, (2) growth, (3) MRDC operation with growth and shipment via ocean starting in month 16, (4) MRDC operation with growth and shipment via air starting in month 16

The implication is that while shipments by air every 2 weeks reduce inventory further than shipments once a month by ocean within a 24 month period the average inventory turns will be essentially the same, with both showing about a 100% improvement over the implementation of MRDC operations.

Yet, inventory turns are not the most indicative representation of improvement of the operation. It is possible to improve inventory turns and make less money as a result. This is indicated in the following graphs which refer to return on inventory assets.

This graph represents resultant LAC return on inventory assets with (1) no growth, (2) growth, (3) MRDC operation with growth and shipment via ocean starting in month 16, (4) MRDC operation with growth and shipment via air starting in month 16

Note that the return on inventory assets for shipments via air is about 5% below the return on inventory assets for shipments by ocean.

This graph represents resultant system (LAC + MRDC) return on inventory assets with (1) no growth, (2) growth, (3) MRDC operation with growth and shipment via ocean starting in month 16, (4) MRDC operation with growth and shipment via air starting in month 16.

The system return on inventory assets is about a percentage point lower than the LAC comparative graph because of the inventory carrying cost for MRDC. Note that the variance between shipping via ocean and air.

Conclusions

From numerous simulation runs with varying input parameter values such as Projected Sales Growth, Forecasting Accuracy, Lead Time, Landing Cost, Service Level, Inventory Carrying Cost, etc. there are several insights that come have become evident. These are provided in the following conclusions.

It is more cost effective to stock at MRDC and ship monthly to the LACs via ocean than to ship every two weeks by air. This option results in a slightly higher inventory maintained in the LACs than if shipments are done every two weeks by air. The combination inventory carrying cost for this higher inventory level combined with the lower landing cost turns out to be more profitable than the combination of inventory carrying cost on a lower level of inventory combined with a higher landing cost. This difference in terms of return on inventory assets is 4 to 5 percent in favor of shipping monthly by ocean.
With the initiation of MRDC operations it is expected that the LAC orders in the pipeline will be routed to MRDC and the LACs will place new orders on a monthly basis with MRDC. These orders will be of significantly lower volume because of the lead time reduction, forecast accuracy improvement and service level improvement.
With the initiation of MRDC operations for all LACs it will take approximately 9 months for the system to stabilize. This stabilization takes into account the amount of time it will take for LAC orders in the pipeline to be delivered to MRDC and the length of time it will take for the LACs to sell down current inventories.
It is recommended that MRDC operation begin one LAC at a time to allow MRDC to develop its operational effectiveness. It is understood that this may not be possible because of the in place forecast / order management system. Whether the transition is done one LAC at a time or all at once I would recommend the preordering of additional inventory for MRDC to cover start up. The absolutely critical portion of this solution is that the LACs believe that MRDC will deliver at a service level higher than the standard suppliers. If the LACs are not convinced of this they will continue create oscillations in the system by over, and later under, ordering. The only way they will come to believe in the new MRDC service level is for MRDC to deliver! Over time the MRDC inventory could be slowly reduced down to more appropriate levels.
Once MRDC has converted all LACs to source from MRDC the next portion of the situation which should be worked on is the reduction of lead times from MRDC suppliers, supplier service levels, and the improvement of forecasting accuracy. These factors are components of safety stock computations and will result in a reduction of the inventory carried by MRDC, thus reducing the inventory carrying cost and improving the overall system return on inventory assets.
The next stage is to work on the lead times and forecasting accuracy for LACs. By improving the forecasting accuracy and reducing the lead time the LACs inventory level my be further reduced improving both the LACs and System return on inventory assets. A key portion of this is finding a way to reduce the shipping cost by air. This is the major determinant of the Cost of Goods Sold which results in a reduced return on inventory assets with shipments by air.
Once this stage is completed the MRDC operation should be concentrated on again to reduce inventory carrying cost, increase forecasting accuracy, reduce lead times to LACs, and improve service level.

This investigation was done in ithink 3.0. The ithink model can be downloaded by clicking here [idm01.zip, 42k].